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);
45 # Setup file and tag signatures. These should never change.
47 sub SIG_FILE () { 'DPDB' }
48 sub SIG_HASH () { 'H' }
49 sub SIG_ARRAY () { 'A' }
50 sub SIG_SCALAR () { 'S' }
51 sub SIG_NULL () { 'N' }
52 sub SIG_DATA () { 'D' }
53 sub SIG_INDEX () { 'I' }
54 sub SIG_BLIST () { 'B' }
58 # Setup constants for users to pass to new()
60 sub TYPE_HASH () { SIG_HASH }
61 sub TYPE_ARRAY () { SIG_ARRAY }
62 sub TYPE_SCALAR () { SIG_SCALAR }
70 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
75 unless ( eval { local $SIG{'__DIE__'}; %{$_[0]} || 1 } ) {
76 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
81 $args = { file => shift };
89 # Class constructor method for Perl OO interface.
90 # Calls tie() and returns blessed reference to tied hash or array,
91 # providing a hybrid OO/tie interface.
94 my $args = $class->_get_args( @_ );
97 # Check if we want a tied hash or array.
100 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
101 $class = 'DBM::Deep::Array';
102 require DBM::Deep::Array;
103 tie @$self, $class, %$args;
106 $class = 'DBM::Deep::Hash';
107 require DBM::Deep::Hash;
108 tie %$self, $class, %$args;
111 return bless $self, $class;
116 # Setup $self and bless into this class.
121 # These are the defaults to be optionally overridden below
124 base_offset => length(SIG_FILE),
125 engine => 'DBM::Deep::Engine',
128 foreach my $param ( keys %$self ) {
129 next unless exists $args->{$param};
130 $self->{$param} = delete $args->{$param}
133 # locking implicitly enables autoflush
134 if ($args->{locking}) { $args->{autoflush} = 1; }
136 $self->{root} = exists $args->{root}
138 : DBM::Deep::_::Root->new( $args );
140 if (!defined($self->_fh)) { $self->{engine}->open( $self ); }
147 require DBM::Deep::Hash;
148 return DBM::Deep::Hash->TIEHASH( @_ );
153 require DBM::Deep::Array;
154 return DBM::Deep::Array->TIEARRAY( @_ );
157 #XXX Unneeded now ...
163 # If db locking is set, flock() the db file. If called multiple
164 # times before unlock(), then the same number of unlocks() must
165 # be called before the lock is released.
167 my $self = $_[0]->_get_self;
169 $type = LOCK_EX unless defined $type;
171 if (!defined($self->_fh)) { return; }
173 if ($self->_root->{locking}) {
174 if (!$self->_root->{locked}) {
175 flock($self->_fh, $type);
177 # refresh end counter in case file has changed size
178 my @stats = stat($self->_root->{file});
179 $self->_root->{end} = $stats[7];
181 # double-check file inode, in case another process
182 # has optimize()d our file while we were waiting.
183 if ($stats[1] != $self->_root->{inode}) {
184 $self->{engine}->open( $self ); # re-open
185 flock($self->_fh, $type); # re-lock
186 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
189 $self->_root->{locked}++;
199 # If db locking is set, unlock the db file. See note in lock()
200 # regarding calling lock() multiple times.
202 my $self = $_[0]->_get_self;
204 if (!defined($self->_fh)) { return; }
206 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
207 $self->_root->{locked}--;
208 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
217 my $self = shift->_get_self;
218 my ($spot, $value) = @_;
223 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
224 my $type = $value->_type;
225 ${$spot} = $type eq TYPE_HASH ? {} : [];
226 $value->_copy_node( ${$spot} );
229 my $r = Scalar::Util::reftype( $value );
230 my $c = Scalar::Util::blessed( $value );
231 if ( $r eq 'ARRAY' ) {
232 ${$spot} = [ @{$value} ];
235 ${$spot} = { %{$value} };
237 ${$spot} = bless ${$spot}, $c
246 # Copy single level of keys or elements to new DB handle.
247 # Recurse for nested structures
249 my $self = shift->_get_self;
252 if ($self->_type eq TYPE_HASH) {
253 my $key = $self->first_key();
255 my $value = $self->get($key);
256 $self->_copy_value( \$db_temp->{$key}, $value );
257 $key = $self->next_key($key);
261 my $length = $self->length();
262 for (my $index = 0; $index < $length; $index++) {
263 my $value = $self->get($index);
264 $self->_copy_value( \$db_temp->[$index], $value );
273 # Recursively export into standard Perl hashes and arrays.
275 my $self = $_[0]->_get_self;
278 if ($self->_type eq TYPE_HASH) { $temp = {}; }
279 elsif ($self->_type eq TYPE_ARRAY) { $temp = []; }
282 $self->_copy_node( $temp );
290 # Recursively import Perl hash/array structure
292 #XXX This use of ref() seems to be ok
293 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
295 my $self = $_[0]->_get_self;
298 #XXX This use of ref() seems to be ok
301 # struct is not a reference, so just import based on our type
305 if ($self->_type eq TYPE_HASH) { $struct = {@_}; }
306 elsif ($self->_type eq TYPE_ARRAY) { $struct = [@_]; }
309 my $r = Scalar::Util::reftype($struct) || '';
310 if ($r eq "HASH" && $self->_type eq TYPE_HASH) {
311 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
313 elsif ($r eq "ARRAY" && $self->_type eq TYPE_ARRAY) {
314 $self->push( @$struct );
317 return $self->_throw_error("Cannot import: type mismatch");
325 # Rebuild entire database into new file, then move
326 # it back on top of original.
328 my $self = $_[0]->_get_self;
330 #XXX Need to create a new test for this
331 # if ($self->_root->{links} > 1) {
332 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
335 my $db_temp = DBM::Deep->new(
336 file => $self->_root->{file} . '.tmp',
340 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
344 $self->_copy_node( $db_temp );
348 # Attempt to copy user, group and permissions over to new file
350 my @stats = stat($self->_fh);
351 my $perms = $stats[2] & 07777;
354 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
355 chmod( $perms, $self->_root->{file} . '.tmp' );
357 # q.v. perlport for more information on this variable
358 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
360 # Potential race condition when optmizing on Win32 with locking.
361 # The Windows filesystem requires that the filehandle be closed
362 # before it is overwritten with rename(). This could be redone
366 $self->{engine}->close( $self );
369 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
370 unlink $self->_root->{file} . '.tmp';
372 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
376 $self->{engine}->close( $self );
377 $self->{engine}->open( $self );
384 # Make copy of object and return
386 my $self = $_[0]->_get_self;
388 return DBM::Deep->new(
389 type => $self->_type,
390 base_offset => $self->_base_offset,
396 my %is_legal_filter = map {
399 store_key store_value
400 fetch_key fetch_value
405 # Setup filter function for storing or fetching the key or value
407 my $self = $_[0]->_get_self;
409 my $func = $_[2] ? $_[2] : undef;
411 if ( $is_legal_filter{$type} ) {
412 $self->_root->{"filter_$type"} = $func;
426 # Get access to the root structure
428 my $self = $_[0]->_get_self;
429 return $self->{root};
434 # Get access to the raw fh
436 #XXX It will be useful, though, when we split out HASH and ARRAY
437 my $self = $_[0]->_get_self;
438 return $self->_root->{fh};
443 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
445 my $self = $_[0]->_get_self;
446 return $self->{type};
451 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
453 my $self = $_[0]->_get_self;
454 return $self->{base_offset};
462 die "DBM::Deep: $_[1]\n";
467 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
472 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
476 # tie() methods (hashes and arrays)
481 # Store single hash key/value or array element in database.
483 my $self = $_[0]->_get_self;
486 # User may be storing a hash, in which case we do not want it run
487 # through the filtering system
488 my $value = ($self->_root->{filter_store_value} && !ref($_[2]))
489 ? $self->_root->{filter_store_value}->($_[2])
492 my $md5 = $DBM::Deep::Engine::DIGEST_FUNC->($key);
494 unless ( _is_writable( $self->_fh ) ) {
495 $self->_throw_error( 'Cannot write to a readonly filehandle' );
499 # Request exclusive lock for writing
501 $self->lock( LOCK_EX );
506 # Locate offset for bucket list using digest index system
508 my $tag = $self->{engine}->load_tag($self, $self->_base_offset);
510 $tag = $self->{engine}->create_tag($self, $self->_base_offset, SIG_INDEX, chr(0) x $DBM::Deep::Engine::INDEX_SIZE);
514 while ($tag->{signature} ne SIG_BLIST) {
515 my $num = ord(substr($md5, $ch, 1));
517 my $ref_loc = $tag->{offset} + ($num * $DBM::Deep::Engine::LONG_SIZE);
518 my $new_tag = $self->{engine}->index_lookup($self, $tag, $num);
521 seek($fh, $ref_loc + $self->_root->{file_offset}, SEEK_SET);
522 print( $fh pack($DBM::Deep::Engine::LONG_PACK, $self->_root->{end}) );
524 $tag = $self->{engine}->create_tag($self, $self->_root->{end}, SIG_BLIST, chr(0) x $DBM::Deep::Engine::BUCKET_LIST_SIZE);
526 $tag->{ref_loc} = $ref_loc;
534 $tag->{ref_loc} = $ref_loc;
541 # Add key/value to bucket list
543 my $result = $self->{engine}->add_bucket( $self, $tag, $md5, $key, $value );
552 # Fetch single value or element given plain key or array index
554 my $self = shift->_get_self;
557 my $md5 = $DBM::Deep::Engine::DIGEST_FUNC->($key);
560 # Request shared lock for reading
562 $self->lock( LOCK_SH );
564 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
571 # Get value from bucket list
573 my $result = $self->{engine}->get_bucket_value( $self, $tag, $md5 );
577 #XXX What is ref() checking here?
578 #YYY Filters only apply on scalar values, so the ref check is making
579 #YYY sure the fetched bucket is a scalar, not a child hash or array.
580 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
581 ? $self->_root->{filter_fetch_value}->($result)
587 # Delete single key/value pair or element given plain key or array index
589 my $self = $_[0]->_get_self;
592 my $md5 = $DBM::Deep::Engine::DIGEST_FUNC->($key);
595 # Request exclusive lock for writing
597 $self->lock( LOCK_EX );
599 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
608 my $value = $self->{engine}->get_bucket_value($self, $tag, $md5 );
609 if ($value && !ref($value) && $self->_root->{filter_fetch_value}) {
610 $value = $self->_root->{filter_fetch_value}->($value);
613 my $result = $self->{engine}->delete_bucket( $self, $tag, $md5 );
616 # If this object is an array and the key deleted was on the end of the stack,
617 # decrement the length variable.
627 # Check if a single key or element exists given plain key or array index
629 my $self = $_[0]->_get_self;
632 my $md5 = $DBM::Deep::Engine::DIGEST_FUNC->($key);
635 # Request shared lock for reading
637 $self->lock( LOCK_SH );
639 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
642 # For some reason, the built-in exists() function returns '' for false
650 # Check if bucket exists and return 1 or ''
652 my $result = $self->{engine}->bucket_exists( $self, $tag, $md5 ) || '';
661 # Clear all keys from hash, or all elements from array.
663 my $self = $_[0]->_get_self;
666 # Request exclusive lock for writing
668 $self->lock( LOCK_EX );
672 seek($fh, $self->_base_offset + $self->_root->{file_offset}, SEEK_SET);
678 $self->{engine}->create_tag($self, $self->_base_offset, $self->_type, chr(0) x $DBM::Deep::Engine::INDEX_SIZE);
686 # Public method aliases
688 sub put { (shift)->STORE( @_ ) }
689 sub store { (shift)->STORE( @_ ) }
690 sub get { (shift)->FETCH( @_ ) }
691 sub fetch { (shift)->FETCH( @_ ) }
692 sub delete { (shift)->DELETE( @_ ) }
693 sub exists { (shift)->EXISTS( @_ ) }
694 sub clear { (shift)->CLEAR( @_ ) }
696 package DBM::Deep::_::Root;
710 filter_store_key => undef,
711 filter_store_value => undef,
712 filter_fetch_key => undef,
713 filter_fetch_value => undef,
718 if ( $self->{fh} && !$self->{file_offset} ) {
719 $self->{file_offset} = tell( $self->{fh} );
729 close $self->{fh} if $self->{fh};
740 DBM::Deep - A pure perl multi-level hash/array DBM
745 my $db = DBM::Deep->new( "foo.db" );
747 $db->{key} = 'value'; # tie() style
750 $db->put('key' => 'value'); # OO style
751 print $db->get('key');
753 # true multi-level support
754 $db->{my_complex} = [
755 'hello', { perl => 'rules' },
761 A unique flat-file database module, written in pure perl. True
762 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
763 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
764 handle millions of keys and unlimited hash levels without significant
765 slow-down. Written from the ground-up in pure perl -- this is NOT a
766 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
767 Mac OS X and Windows.
771 Hopefully you are using Perl's excellent CPAN module, which will download
772 and install the module for you. If not, get the tarball, and run these
784 Construction can be done OO-style (which is the recommended way), or using
785 Perl's tie() function. Both are examined here.
787 =head2 OO CONSTRUCTION
789 The recommended way to construct a DBM::Deep object is to use the new()
790 method, which gets you a blessed, tied hash or array reference.
792 my $db = DBM::Deep->new( "foo.db" );
794 This opens a new database handle, mapped to the file "foo.db". If this
795 file does not exist, it will automatically be created. DB files are
796 opened in "r+" (read/write) mode, and the type of object returned is a
797 hash, unless otherwise specified (see L<OPTIONS> below).
799 You can pass a number of options to the constructor to specify things like
800 locking, autoflush, etc. This is done by passing an inline hash:
802 my $db = DBM::Deep->new(
808 Notice that the filename is now specified I<inside> the hash with
809 the "file" parameter, as opposed to being the sole argument to the
810 constructor. This is required if any options are specified.
811 See L<OPTIONS> below for the complete list.
815 You can also start with an array instead of a hash. For this, you must
816 specify the C<type> parameter:
818 my $db = DBM::Deep->new(
820 type => DBM::Deep->TYPE_ARRAY
823 B<Note:> Specifing the C<type> parameter only takes effect when beginning
824 a new DB file. If you create a DBM::Deep object with an existing file, the
825 C<type> will be loaded from the file header, and an error will be thrown if
826 the wrong type is passed in.
828 =head2 TIE CONSTRUCTION
830 Alternately, you can create a DBM::Deep handle by using Perl's built-in
831 tie() function. The object returned from tie() can be used to call methods,
832 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
833 file (as expected with most tie'd objects).
836 my $db = tie %hash, "DBM::Deep", "foo.db";
839 my $db = tie @array, "DBM::Deep", "bar.db";
841 As with the OO constructor, you can replace the DB filename parameter with
842 a hash containing one or more options (see L<OPTIONS> just below for the
845 tie %hash, "DBM::Deep", {
853 There are a number of options that can be passed in when constructing your
854 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
860 Filename of the DB file to link the handle to. You can pass a full absolute
861 filesystem path, partial path, or a plain filename if the file is in the
862 current working directory. This is a required parameter (though q.v. fh).
866 If you want, you can pass in the fh instead of the file. This is most useful for doing
869 my $db = DBM::Deep->new( { fh => \*DATA } );
871 You are responsible for making sure that the fh has been opened appropriately for your
872 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
873 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
874 needs to read from the fh.
878 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
879 not need to set this. However, it's there if you want it.
881 If you pass in fh and do not set this, it will be set appropriately.
885 This parameter specifies what type of object to create, a hash or array. Use
886 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
887 This only takes effect when beginning a new file. This is an optional
888 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
892 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
893 function to lock the database in exclusive mode for writes, and shared mode for
894 reads. Pass any true value to enable. This affects the base DB handle I<and
895 any child hashes or arrays> that use the same DB file. This is an optional
896 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
900 Specifies whether autoflush is to be enabled on the underlying filehandle.
901 This obviously slows down write operations, but is required if you may have
902 multiple processes accessing the same DB file (also consider enable I<locking>).
903 Pass any true value to enable. This is an optional parameter, and defaults to 0
908 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
909 restore them when fetched. This is an B<experimental> feature, and does have
910 side-effects. Basically, when hashes are re-blessed into their original
911 classes, they are no longer blessed into the DBM::Deep class! So you won't be
912 able to call any DBM::Deep methods on them. You have been warned.
913 This is an optional parameter, and defaults to 0 (disabled).
917 See L<FILTERS> below.
923 With DBM::Deep you can access your databases using Perl's standard hash/array
924 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
925 treat them as such. DBM::Deep will intercept all reads/writes and direct them
926 to the right place -- the DB file. This has nothing to do with the
927 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
928 using regular hashes and arrays, rather than calling functions like C<get()>
929 and C<put()> (although those work too). It is entirely up to you how to want
930 to access your databases.
934 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
935 or even nested hashes (or arrays) using standard Perl syntax:
937 my $db = DBM::Deep->new( "foo.db" );
939 $db->{mykey} = "myvalue";
941 $db->{myhash}->{subkey} = "subvalue";
943 print $db->{myhash}->{subkey} . "\n";
945 You can even step through hash keys using the normal Perl C<keys()> function:
947 foreach my $key (keys %$db) {
948 print "$key: " . $db->{$key} . "\n";
951 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
952 pushes them onto an array, all before the loop even begins. If you have an
953 extra large hash, this may exhaust Perl's memory. Instead, consider using
954 Perl's C<each()> function, which pulls keys/values one at a time, using very
957 while (my ($key, $value) = each %$db) {
958 print "$key: $value\n";
961 Please note that when using C<each()>, you should always pass a direct
962 hash reference, not a lookup. Meaning, you should B<never> do this:
965 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
967 This causes an infinite loop, because for each iteration, Perl is calling
968 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
969 it effectively keeps returning the first key over and over again. Instead,
970 assign a temporary variable to C<$db->{foo}>, then pass that to each().
974 As with hashes, you can treat any DBM::Deep object like a normal Perl array
975 reference. This includes inserting, removing and manipulating elements,
976 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
977 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
978 or simply be a nested array reference inside a hash. Example:
980 my $db = DBM::Deep->new(
981 file => "foo-array.db",
982 type => DBM::Deep->TYPE_ARRAY
986 push @$db, "bar", "baz";
989 my $last_elem = pop @$db; # baz
990 my $first_elem = shift @$db; # bah
991 my $second_elem = $db->[1]; # bar
993 my $num_elements = scalar @$db;
997 In addition to the I<tie()> interface, you can also use a standard OO interface
998 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
999 array) has its own methods, but both types share the following common methods:
1000 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1004 =item * new() / clone()
1006 These are the constructor and copy-functions.
1008 =item * put() / store()
1010 Stores a new hash key/value pair, or sets an array element value. Takes two
1011 arguments, the hash key or array index, and the new value. The value can be
1012 a scalar, hash ref or array ref. Returns true on success, false on failure.
1014 $db->put("foo", "bar"); # for hashes
1015 $db->put(1, "bar"); # for arrays
1017 =item * get() / fetch()
1019 Fetches the value of a hash key or array element. Takes one argument: the hash
1020 key or array index. Returns a scalar, hash ref or array ref, depending on the
1023 my $value = $db->get("foo"); # for hashes
1024 my $value = $db->get(1); # for arrays
1028 Checks if a hash key or array index exists. Takes one argument: the hash key
1029 or array index. Returns true if it exists, false if not.
1031 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1032 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1036 Deletes one hash key/value pair or array element. Takes one argument: the hash
1037 key or array index. Returns true on success, false if not found. For arrays,
1038 the remaining elements located after the deleted element are NOT moved over.
1039 The deleted element is essentially just undefined, which is exactly how Perl's
1040 internal arrays work. Please note that the space occupied by the deleted
1041 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1042 below for details and workarounds.
1044 $db->delete("foo"); # for hashes
1045 $db->delete(1); # for arrays
1049 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1050 value. Please note that the space occupied by the deleted keys/values or
1051 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1052 details and workarounds.
1054 $db->clear(); # hashes or arrays
1056 =item * lock() / unlock()
1062 Recover lost disk space.
1064 =item * import() / export()
1066 Data going in and out.
1068 =item * set_digest() / set_pack() / set_filter()
1070 q.v. adjusting the interal parameters.
1076 For hashes, DBM::Deep supports all the common methods described above, and the
1077 following additional methods: C<first_key()> and C<next_key()>.
1083 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1084 fetched in an undefined order (which appears random). Takes no arguments,
1085 returns the key as a scalar value.
1087 my $key = $db->first_key();
1091 Returns the "next" key in the hash, given the previous one as the sole argument.
1092 Returns undef if there are no more keys to be fetched.
1094 $key = $db->next_key($key);
1098 Here are some examples of using hashes:
1100 my $db = DBM::Deep->new( "foo.db" );
1102 $db->put("foo", "bar");
1103 print "foo: " . $db->get("foo") . "\n";
1105 $db->put("baz", {}); # new child hash ref
1106 $db->get("baz")->put("buz", "biz");
1107 print "buz: " . $db->get("baz")->get("buz") . "\n";
1109 my $key = $db->first_key();
1111 print "$key: " . $db->get($key) . "\n";
1112 $key = $db->next_key($key);
1115 if ($db->exists("foo")) { $db->delete("foo"); }
1119 For arrays, DBM::Deep supports all the common methods described above, and the
1120 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1121 C<unshift()> and C<splice()>.
1127 Returns the number of elements in the array. Takes no arguments.
1129 my $len = $db->length();
1133 Adds one or more elements onto the end of the array. Accepts scalars, hash
1134 refs or array refs. No return value.
1136 $db->push("foo", "bar", {});
1140 Fetches the last element in the array, and deletes it. Takes no arguments.
1141 Returns undef if array is empty. Returns the element value.
1143 my $elem = $db->pop();
1147 Fetches the first element in the array, deletes it, then shifts all the
1148 remaining elements over to take up the space. Returns the element value. This
1149 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1152 my $elem = $db->shift();
1156 Inserts one or more elements onto the beginning of the array, shifting all
1157 existing elements over to make room. Accepts scalars, hash refs or array refs.
1158 No return value. This method is not recommended with large arrays -- see
1159 <LARGE ARRAYS> below for details.
1161 $db->unshift("foo", "bar", {});
1165 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1166 -f splice> for usage -- it is too complicated to document here. This method is
1167 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1171 Here are some examples of using arrays:
1173 my $db = DBM::Deep->new(
1175 type => DBM::Deep->TYPE_ARRAY
1178 $db->push("bar", "baz");
1179 $db->unshift("foo");
1182 my $len = $db->length();
1183 print "length: $len\n"; # 4
1185 for (my $k=0; $k<$len; $k++) {
1186 print "$k: " . $db->get($k) . "\n";
1189 $db->splice(1, 2, "biz", "baf");
1191 while (my $elem = shift @$db) {
1192 print "shifted: $elem\n";
1197 Enable automatic file locking by passing a true value to the C<locking>
1198 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1200 my $db = DBM::Deep->new(
1205 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1206 mode for writes, and shared mode for reads. This is required if you have
1207 multiple processes accessing the same database file, to avoid file corruption.
1208 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1209 NFS> below for more.
1211 =head2 EXPLICIT LOCKING
1213 You can explicitly lock a database, so it remains locked for multiple
1214 transactions. This is done by calling the C<lock()> method, and passing an
1215 optional lock mode argument (defaults to exclusive mode). This is particularly
1216 useful for things like counters, where the current value needs to be fetched,
1217 then incremented, then stored again.
1220 my $counter = $db->get("counter");
1222 $db->put("counter", $counter);
1231 You can pass C<lock()> an optional argument, which specifies which mode to use
1232 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1233 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1234 same as the constants defined in Perl's C<Fcntl> module.
1236 $db->lock( DBM::Deep->LOCK_SH );
1240 =head1 IMPORTING/EXPORTING
1242 You can import existing complex structures by calling the C<import()> method,
1243 and export an entire database into an in-memory structure using the C<export()>
1244 method. Both are examined here.
1248 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1249 walking the structure and adding keys/elements to the database as you go,
1250 simply pass a reference to the C<import()> method. This recursively adds
1251 everything to an existing DBM::Deep object for you. Here is an example:
1256 array1 => [ "elem0", "elem1", "elem2" ],
1258 subkey1 => "subvalue1",
1259 subkey2 => "subvalue2"
1263 my $db = DBM::Deep->new( "foo.db" );
1264 $db->import( $struct );
1266 print $db->{key1} . "\n"; # prints "value1"
1268 This recursively imports the entire C<$struct> object into C<$db>, including
1269 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1270 keys are merged with the existing ones, replacing if they already exist.
1271 The C<import()> method can be called on any database level (not just the base
1272 level), and works with both hash and array DB types.
1274 B<Note:> Make sure your existing structure has no circular references in it.
1275 These will cause an infinite loop when importing.
1279 Calling the C<export()> method on an existing DBM::Deep object will return
1280 a reference to a new in-memory copy of the database. The export is done
1281 recursively, so all nested hashes/arrays are all exported to standard Perl
1282 objects. Here is an example:
1284 my $db = DBM::Deep->new( "foo.db" );
1286 $db->{key1} = "value1";
1287 $db->{key2} = "value2";
1289 $db->{hash1}->{subkey1} = "subvalue1";
1290 $db->{hash1}->{subkey2} = "subvalue2";
1292 my $struct = $db->export();
1294 print $struct->{key1} . "\n"; # prints "value1"
1296 This makes a complete copy of the database in memory, and returns a reference
1297 to it. The C<export()> method can be called on any database level (not just
1298 the base level), and works with both hash and array DB types. Be careful of
1299 large databases -- you can store a lot more data in a DBM::Deep object than an
1300 in-memory Perl structure.
1302 B<Note:> Make sure your database has no circular references in it.
1303 These will cause an infinite loop when exporting.
1307 DBM::Deep has a number of hooks where you can specify your own Perl function
1308 to perform filtering on incoming or outgoing data. This is a perfect
1309 way to extend the engine, and implement things like real-time compression or
1310 encryption. Filtering applies to the base DB level, and all child hashes /
1311 arrays. Filter hooks can be specified when your DBM::Deep object is first
1312 constructed, or by calling the C<set_filter()> method at any time. There are
1313 four available filter hooks, described below:
1317 =item * filter_store_key
1319 This filter is called whenever a hash key is stored. It
1320 is passed the incoming key, and expected to return a transformed key.
1322 =item * filter_store_value
1324 This filter is called whenever a hash key or array element is stored. It
1325 is passed the incoming value, and expected to return a transformed value.
1327 =item * filter_fetch_key
1329 This filter is called whenever a hash key is fetched (i.e. via
1330 C<first_key()> or C<next_key()>). It is passed the transformed key,
1331 and expected to return the plain key.
1333 =item * filter_fetch_value
1335 This filter is called whenever a hash key or array element is fetched.
1336 It is passed the transformed value, and expected to return the plain value.
1340 Here are the two ways to setup a filter hook:
1342 my $db = DBM::Deep->new(
1344 filter_store_value => \&my_filter_store,
1345 filter_fetch_value => \&my_filter_fetch
1350 $db->set_filter( "filter_store_value", \&my_filter_store );
1351 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1353 Your filter function will be called only when dealing with SCALAR keys or
1354 values. When nested hashes and arrays are being stored/fetched, filtering
1355 is bypassed. Filters are called as static functions, passed a single SCALAR
1356 argument, and expected to return a single SCALAR value. If you want to
1357 remove a filter, set the function reference to C<undef>:
1359 $db->set_filter( "filter_store_value", undef );
1361 =head2 REAL-TIME ENCRYPTION EXAMPLE
1363 Here is a working example that uses the I<Crypt::Blowfish> module to
1364 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1365 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1366 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1369 use Crypt::Blowfish;
1372 my $cipher = Crypt::CBC->new({
1373 'key' => 'my secret key',
1374 'cipher' => 'Blowfish',
1376 'regenerate_key' => 0,
1377 'padding' => 'space',
1381 my $db = DBM::Deep->new(
1382 file => "foo-encrypt.db",
1383 filter_store_key => \&my_encrypt,
1384 filter_store_value => \&my_encrypt,
1385 filter_fetch_key => \&my_decrypt,
1386 filter_fetch_value => \&my_decrypt,
1389 $db->{key1} = "value1";
1390 $db->{key2} = "value2";
1391 print "key1: " . $db->{key1} . "\n";
1392 print "key2: " . $db->{key2} . "\n";
1398 return $cipher->encrypt( $_[0] );
1401 return $cipher->decrypt( $_[0] );
1404 =head2 REAL-TIME COMPRESSION EXAMPLE
1406 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1407 compression / decompression of keys & values with DBM::Deep Filters.
1408 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1409 more on I<Compress::Zlib>.
1414 my $db = DBM::Deep->new(
1415 file => "foo-compress.db",
1416 filter_store_key => \&my_compress,
1417 filter_store_value => \&my_compress,
1418 filter_fetch_key => \&my_decompress,
1419 filter_fetch_value => \&my_decompress,
1422 $db->{key1} = "value1";
1423 $db->{key2} = "value2";
1424 print "key1: " . $db->{key1} . "\n";
1425 print "key2: " . $db->{key2} . "\n";
1431 return Compress::Zlib::memGzip( $_[0] ) ;
1434 return Compress::Zlib::memGunzip( $_[0] ) ;
1437 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1438 actually numerical index numbers, and are not filtered.
1440 =head1 ERROR HANDLING
1442 Most DBM::Deep methods return a true value for success, and call die() on
1443 failure. You can wrap calls in an eval block to catch the die.
1445 my $db = DBM::Deep->new( "foo.db" ); # create hash
1446 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1448 print $@; # prints error message
1450 =head1 LARGEFILE SUPPORT
1452 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1453 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1454 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1455 by calling the static C<set_pack()> method before you do anything else.
1457 DBM::Deep::set_pack(8, 'Q');
1459 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1460 instead of 32-bit longs. After setting these values your DB files have a
1461 theoretical maximum size of 16 XB (exabytes).
1463 B<Note:> Changing these values will B<NOT> work for existing database files.
1464 Only change this for new files, and make sure it stays set consistently
1465 throughout the file's life. If you do set these values, you can no longer
1466 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
1467 back to 32-bit mode.
1469 B<Note:> I have not personally tested files > 2 GB -- all my systems have
1470 only a 32-bit Perl. However, I have received user reports that this does
1473 =head1 LOW-LEVEL ACCESS
1475 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1476 you can call the C<_fh()> method, which returns the handle:
1478 my $fh = $db->_fh();
1480 This method can be called on the root level of the datbase, or any child
1481 hashes or arrays. All levels share a I<root> structure, which contains things
1482 like the filehandle, a reference counter, and all the options specified
1483 when you created the object. You can get access to this root structure by
1484 calling the C<root()> method.
1486 my $root = $db->_root();
1488 This is useful for changing options after the object has already been created,
1489 such as enabling/disabling locking. You can also store your own temporary user
1490 data in this structure (be wary of name collision), which is then accessible from
1491 any child hash or array.
1493 =head1 CUSTOM DIGEST ALGORITHM
1495 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1496 keys. However you can override this, and use another algorithm (such as SHA-256)
1497 or even write your own. But please note that DBM::Deep currently expects zero
1498 collisions, so your algorithm has to be I<perfect>, so to speak.
1499 Collision detection may be introduced in a later version.
1503 You can specify a custom digest algorithm by calling the static C<set_digest()>
1504 function, passing a reference to a subroutine, and the length of the algorithm's
1505 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
1506 objects. Here is a working example that uses a 256-bit hash from the
1507 I<Digest::SHA256> module. Please see
1508 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
1513 my $context = Digest::SHA256::new(256);
1515 DBM::Deep::set_digest( \&my_digest, 32 );
1517 my $db = DBM::Deep->new( "foo-sha.db" );
1519 $db->{key1} = "value1";
1520 $db->{key2} = "value2";
1521 print "key1: " . $db->{key1} . "\n";
1522 print "key2: " . $db->{key2} . "\n";
1528 return substr( $context->hash($_[0]), 0, 32 );
1531 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1532 of bytes you specify in the C<set_digest()> function (in this case 32).
1534 =head1 CIRCULAR REFERENCES
1536 DBM::Deep has B<experimental> support for circular references. Meaning you
1537 can have a nested hash key or array element that points to a parent object.
1538 This relationship is stored in the DB file, and is preserved between sessions.
1541 my $db = DBM::Deep->new( "foo.db" );
1544 $db->{circle} = $db; # ref to self
1546 print $db->{foo} . "\n"; # prints "foo"
1547 print $db->{circle}->{foo} . "\n"; # prints "foo" again
1549 One catch is, passing the object to a function that recursively walks the
1550 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1551 C<export()> methods) will result in an infinite loop. The other catch is,
1552 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
1553 or C<next_key()> methods), you will get the I<target object's key>, not the
1554 ref's key. This gets even more interesting with the above example, where
1555 the I<circle> key points to the base DB object, which technically doesn't
1556 have a key. So I made DBM::Deep return "[base]" as the key name in that
1559 =head1 CAVEATS / ISSUES / BUGS
1561 This section describes all the known issues with DBM::Deep. It you have found
1562 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1564 =head2 UNUSED SPACE RECOVERY
1566 One major caveat with DBM::Deep is that space occupied by existing keys and
1567 values is not recovered when they are deleted. Meaning if you keep deleting
1568 and adding new keys, your file will continuously grow. I am working on this,
1569 but in the meantime you can call the built-in C<optimize()> method from time to
1570 time (perhaps in a crontab or something) to recover all your unused space.
1572 $db->optimize(); # returns true on success
1574 This rebuilds the ENTIRE database into a new file, then moves it on top of
1575 the original. The new file will have no unused space, thus it will take up as
1576 little disk space as possible. Please note that this operation can take
1577 a long time for large files, and you need enough disk space to temporarily hold
1578 2 copies of your DB file. The temporary file is created in the same directory
1579 as the original, named with a ".tmp" extension, and is deleted when the
1580 operation completes. Oh, and if locking is enabled, the DB is automatically
1581 locked for the entire duration of the copy.
1583 B<WARNING:> Only call optimize() on the top-level node of the database, and
1584 make sure there are no child references lying around. DBM::Deep keeps a reference
1585 counter, and if it is greater than 1, optimize() will abort and return undef.
1587 =head2 AUTOVIVIFICATION
1589 Unfortunately, autovivification doesn't work with tied hashes. This appears to
1590 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
1591 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
1592 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
1593 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
1596 $db->{foo}->{bar} = "hello";
1598 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
1599 being an empty hash. Try this instead, which works fine:
1601 $db->{foo} = { bar => "hello" };
1603 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
1604 the execution path, and Perl indeed passes an empty hash to the STORE() method.
1605 Probably a bug in Perl.
1607 =head2 FILE CORRUPTION
1609 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1610 for a 32-bit signature when opened, but other corruption in files can cause
1611 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1612 stuck in an infinite loop depending on the level of corruption. File write
1613 operations are not checked for failure (for speed), so if you happen to run
1614 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1615 be addressed in a later version of DBM::Deep.
1619 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1620 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1621 about setting up your NFS server with a locking daemon, then using lockf() to
1622 lock your files, but your mileage may vary there as well. From what I
1623 understand, there is no real way to do it. However, if you need access to the
1624 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1625 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1627 =head2 COPYING OBJECTS
1629 Beware of copying tied objects in Perl. Very strange things can happen.
1630 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1631 returns a new, blessed, tied hash or array to the same level in the DB.
1633 my $copy = $db->clone();
1635 B<Note>: Since clone() here is cloning the object, not the database location, any
1636 modifications to either $db or $copy will be visible in both.
1640 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1641 These functions cause every element in the array to move, which can be murder
1642 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1643 a different location. This will be addressed in the forthcoming version 1.00.
1645 =head2 WRITEONLY FILES
1647 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1648 writeonly mode. STORE will verify that the filehandle is writable. However, there
1649 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1650 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1654 This section discusses DBM::Deep's speed and memory usage.
1658 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1659 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1660 multi-level hash/array support, and cross-platform FTPable files. Even so,
1661 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1662 with huge databases. Here is some test data:
1664 Adding 1,000,000 keys to new DB file...
1666 At 100 keys, avg. speed is 2,703 keys/sec
1667 At 200 keys, avg. speed is 2,642 keys/sec
1668 At 300 keys, avg. speed is 2,598 keys/sec
1669 At 400 keys, avg. speed is 2,578 keys/sec
1670 At 500 keys, avg. speed is 2,722 keys/sec
1671 At 600 keys, avg. speed is 2,628 keys/sec
1672 At 700 keys, avg. speed is 2,700 keys/sec
1673 At 800 keys, avg. speed is 2,607 keys/sec
1674 At 900 keys, avg. speed is 2,190 keys/sec
1675 At 1,000 keys, avg. speed is 2,570 keys/sec
1676 At 2,000 keys, avg. speed is 2,417 keys/sec
1677 At 3,000 keys, avg. speed is 1,982 keys/sec
1678 At 4,000 keys, avg. speed is 1,568 keys/sec
1679 At 5,000 keys, avg. speed is 1,533 keys/sec
1680 At 6,000 keys, avg. speed is 1,787 keys/sec
1681 At 7,000 keys, avg. speed is 1,977 keys/sec
1682 At 8,000 keys, avg. speed is 2,028 keys/sec
1683 At 9,000 keys, avg. speed is 2,077 keys/sec
1684 At 10,000 keys, avg. speed is 2,031 keys/sec
1685 At 20,000 keys, avg. speed is 1,970 keys/sec
1686 At 30,000 keys, avg. speed is 2,050 keys/sec
1687 At 40,000 keys, avg. speed is 2,073 keys/sec
1688 At 50,000 keys, avg. speed is 1,973 keys/sec
1689 At 60,000 keys, avg. speed is 1,914 keys/sec
1690 At 70,000 keys, avg. speed is 2,091 keys/sec
1691 At 80,000 keys, avg. speed is 2,103 keys/sec
1692 At 90,000 keys, avg. speed is 1,886 keys/sec
1693 At 100,000 keys, avg. speed is 1,970 keys/sec
1694 At 200,000 keys, avg. speed is 2,053 keys/sec
1695 At 300,000 keys, avg. speed is 1,697 keys/sec
1696 At 400,000 keys, avg. speed is 1,838 keys/sec
1697 At 500,000 keys, avg. speed is 1,941 keys/sec
1698 At 600,000 keys, avg. speed is 1,930 keys/sec
1699 At 700,000 keys, avg. speed is 1,735 keys/sec
1700 At 800,000 keys, avg. speed is 1,795 keys/sec
1701 At 900,000 keys, avg. speed is 1,221 keys/sec
1702 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1704 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1705 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1706 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1707 Run time was 12 min 3 sec.
1711 One of the great things about DBM::Deep is that it uses very little memory.
1712 Even with huge databases (1,000,000+ keys) you will not see much increased
1713 memory on your process. DBM::Deep relies solely on the filesystem for storing
1714 and fetching data. Here is output from I</usr/bin/top> before even opening a
1717 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1718 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1720 Basically the process is taking 2,716K of memory. And here is the same
1721 process after storing and fetching 1,000,000 keys:
1723 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1724 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1726 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1727 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1729 =head1 DB FILE FORMAT
1731 In case you were interested in the underlying DB file format, it is documented
1732 here in this section. You don't need to know this to use the module, it's just
1733 included for reference.
1737 DBM::Deep files always start with a 32-bit signature to identify the file type.
1738 This is at offset 0. The signature is "DPDB" in network byte order. This is
1739 checked for when the file is opened and an error will be thrown if it's not found.
1743 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
1744 has a standard header containing the type of data, the length of data, and then
1745 the data itself. The type is a single character (1 byte), the length is a
1746 32-bit unsigned long in network byte order, and the data is, well, the data.
1747 Here is how it unfolds:
1751 Immediately after the 32-bit file signature is the I<Master Index> record.
1752 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
1753 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
1754 depending on how the DBM::Deep object was constructed.
1756 The index works by looking at a I<MD5 Hash> of the hash key (or array index
1757 number). The first 8-bit char of the MD5 signature is the offset into the
1758 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
1759 index element is a file offset of the next tag for the key/element in question,
1760 which is usually a I<Bucket List> tag (see below).
1762 The next tag I<could> be another index, depending on how many keys/elements
1763 exist. See L<RE-INDEXING> below for details.
1767 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
1768 file offsets to where the actual data is stored. It starts with a standard
1769 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
1770 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
1771 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
1772 When the list fills up, a I<Re-Index> operation is performed (See
1773 L<RE-INDEXING> below).
1777 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
1778 index/value pair (in array mode). It starts with a standard tag header with
1779 type I<D> for scalar data (string, binary, etc.), or it could be a nested
1780 hash (type I<H>) or array (type I<A>). The value comes just after the tag
1781 header. The size reported in the tag header is only for the value, but then,
1782 just after the value is another size (32-bit unsigned long) and then the plain
1783 key itself. Since the value is likely to be fetched more often than the plain
1784 key, I figured it would be I<slightly> faster to store the value first.
1786 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
1787 record for the nested structure, where the process begins all over again.
1791 After a I<Bucket List> grows to 16 records, its allocated space in the file is
1792 exhausted. Then, when another key/element comes in, the list is converted to a
1793 new index record. However, this index will look at the next char in the MD5
1794 hash, and arrange new Bucket List pointers accordingly. This process is called
1795 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
1796 17 (16 + new one) keys/elements are removed from the old Bucket List and
1797 inserted into the new index. Several new Bucket Lists are created in the
1798 process, as a new MD5 char from the key is being examined (it is unlikely that
1799 the keys will all share the same next char of their MD5s).
1801 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
1802 when the Bucket Lists will turn into indexes, but the first round tends to
1803 happen right around 4,000 keys. You will see a I<slight> decrease in
1804 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
1805 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
1806 right around 900,000 keys. This process can continue nearly indefinitely --
1807 right up until the point the I<MD5> signatures start colliding with each other,
1808 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
1809 getting struck by lightning while you are walking to cash in your tickets.
1810 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
1811 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
1812 this is 340 unodecillion, but don't quote me).
1816 When a new key/element is stored, the key (or index number) is first run through
1817 I<Digest::MD5> to get a 128-bit signature (example, in hex:
1818 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
1819 for the first char of the signature (in this case I<b0>). If it does not exist,
1820 a new I<Bucket List> is created for our key (and the next 15 future keys that
1821 happen to also have I<b> as their first MD5 char). The entire MD5 is written
1822 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
1823 this point, unless we are replacing an existing I<Bucket>), where the actual
1824 data will be stored.
1828 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
1829 (or index number), then walking along the indexes. If there are enough
1830 keys/elements in this DB level, there might be nested indexes, each linked to
1831 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
1832 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
1833 question. If we found a match, the I<Bucket> tag is loaded, where the value and
1834 plain key are stored.
1836 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
1837 methods. In this process the indexes are walked systematically, and each key
1838 fetched in increasing MD5 order (which is why it appears random). Once the
1839 I<Bucket> is found, the value is skipped and the plain key returned instead.
1840 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
1841 alphabetically sorted. This only happens on an index-level -- as soon as the
1842 I<Bucket Lists> are hit, the keys will come out in the order they went in --
1843 so it's pretty much undefined how the keys will come out -- just like Perl's
1846 =head1 CODE COVERAGE
1848 We use B<Devel::Cover> to test the code coverage of our tests, below is the
1849 B<Devel::Cover> report on this module's test suite.
1851 ---------------------------- ------ ------ ------ ------ ------ ------ ------
1852 File stmt bran cond sub pod time total
1853 ---------------------------- ------ ------ ------ ------ ------ ------ ------
1854 blib/lib/DBM/Deep.pm 95.2 83.8 70.0 98.2 100.0 58.0 91.0
1855 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 n/a 26.7 98.0
1856 blib/lib/DBM/Deep/Hash.pm 95.3 80.0 100.0 100.0 n/a 15.3 92.4
1857 Total 96.2 84.8 74.4 98.8 100.0 100.0 92.4
1858 ---------------------------- ------ ------ ------ ------ ------ ------ ------
1860 =head1 MORE INFORMATION
1862 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
1863 or send email to L<DBM-Deep@googlegroups.com>.
1867 Joseph Huckaby, L<jhuckaby@cpan.org>
1869 Rob Kinyon, L<rkinyon@cpan.org>
1871 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
1875 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
1876 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
1880 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
1881 This is free software, you may use it and distribute it under the
1882 same terms as Perl itself.