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->new,
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 $self->{engine}->setup_fh( $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}->close( $self );
185 $self->{engine}->setup_fh( $self );
186 flock($self->_fh, $type); # re-lock
188 # This may not be necessary after re-opening
189 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
192 $self->_root->{locked}++;
202 # If db locking is set, unlock the db file. See note in lock()
203 # regarding calling lock() multiple times.
205 my $self = $_[0]->_get_self;
207 if (!defined($self->_fh)) { return; }
209 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
210 $self->_root->{locked}--;
211 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
220 my $self = shift->_get_self;
221 my ($spot, $value) = @_;
226 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
227 my $type = $value->_type;
228 ${$spot} = $type eq TYPE_HASH ? {} : [];
229 $value->_copy_node( ${$spot} );
232 my $r = Scalar::Util::reftype( $value );
233 my $c = Scalar::Util::blessed( $value );
234 if ( $r eq 'ARRAY' ) {
235 ${$spot} = [ @{$value} ];
238 ${$spot} = { %{$value} };
240 ${$spot} = bless ${$spot}, $c
249 # Copy single level of keys or elements to new DB handle.
250 # Recurse for nested structures
252 my $self = shift->_get_self;
255 if ($self->_type eq TYPE_HASH) {
256 my $key = $self->first_key();
258 my $value = $self->get($key);
259 $self->_copy_value( \$db_temp->{$key}, $value );
260 $key = $self->next_key($key);
264 my $length = $self->length();
265 for (my $index = 0; $index < $length; $index++) {
266 my $value = $self->get($index);
267 $self->_copy_value( \$db_temp->[$index], $value );
276 # Recursively export into standard Perl hashes and arrays.
278 my $self = $_[0]->_get_self;
281 if ($self->_type eq TYPE_HASH) { $temp = {}; }
282 elsif ($self->_type eq TYPE_ARRAY) { $temp = []; }
285 $self->_copy_node( $temp );
293 # Recursively import Perl hash/array structure
295 #XXX This use of ref() seems to be ok
296 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
298 my $self = $_[0]->_get_self;
301 #XXX This use of ref() seems to be ok
304 # struct is not a reference, so just import based on our type
308 if ($self->_type eq TYPE_HASH) { $struct = {@_}; }
309 elsif ($self->_type eq TYPE_ARRAY) { $struct = [@_]; }
312 my $r = Scalar::Util::reftype($struct) || '';
313 if ($r eq "HASH" && $self->_type eq TYPE_HASH) {
314 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
316 elsif ($r eq "ARRAY" && $self->_type eq TYPE_ARRAY) {
317 $self->push( @$struct );
320 return $self->_throw_error("Cannot import: type mismatch");
328 # Rebuild entire database into new file, then move
329 # it back on top of original.
331 my $self = $_[0]->_get_self;
333 #XXX Need to create a new test for this
334 # if ($self->_root->{links} > 1) {
335 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
338 my $db_temp = DBM::Deep->new(
339 file => $self->_root->{file} . '.tmp',
343 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
347 $self->_copy_node( $db_temp );
351 # Attempt to copy user, group and permissions over to new file
353 my @stats = stat($self->_fh);
354 my $perms = $stats[2] & 07777;
357 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
358 chmod( $perms, $self->_root->{file} . '.tmp' );
360 # q.v. perlport for more information on this variable
361 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
363 # Potential race condition when optmizing on Win32 with locking.
364 # The Windows filesystem requires that the filehandle be closed
365 # before it is overwritten with rename(). This could be redone
369 $self->{engine}->close( $self );
372 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
373 unlink $self->_root->{file} . '.tmp';
375 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
379 $self->{engine}->close( $self );
380 $self->{engine}->setup_fh( $self );
387 # Make copy of object and return
389 my $self = $_[0]->_get_self;
391 return DBM::Deep->new(
392 type => $self->_type,
393 base_offset => $self->_base_offset,
399 my %is_legal_filter = map {
402 store_key store_value
403 fetch_key fetch_value
408 # Setup filter function for storing or fetching the key or value
410 my $self = $_[0]->_get_self;
412 my $func = $_[2] ? $_[2] : undef;
414 if ( $is_legal_filter{$type} ) {
415 $self->_root->{"filter_$type"} = $func;
429 # Get access to the root structure
431 my $self = $_[0]->_get_self;
432 return $self->{root};
437 # Get access to the raw fh
439 #XXX It will be useful, though, when we split out HASH and ARRAY
440 my $self = $_[0]->_get_self;
441 return $self->_root->{fh};
446 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
448 my $self = $_[0]->_get_self;
449 return $self->{type};
454 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
456 my $self = $_[0]->_get_self;
457 return $self->{base_offset};
465 die "DBM::Deep: $_[1]\n";
470 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
475 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
479 # tie() methods (hashes and arrays)
484 # Store single hash key/value or array element in database.
486 my $self = $_[0]->_get_self;
489 # User may be storing a hash, in which case we do not want it run
490 # through the filtering system
491 my $value = ($self->_root->{filter_store_value} && !ref($_[2]))
492 ? $self->_root->{filter_store_value}->($_[2])
495 my $md5 = $self->{engine}{digest}->($key);
497 unless ( _is_writable( $self->_fh ) ) {
498 $self->_throw_error( 'Cannot write to a readonly filehandle' );
502 # Request exclusive lock for writing
504 $self->lock( LOCK_EX );
509 # Locate offset for bucket list using digest index system
511 my $tag = $self->{engine}->load_tag($self, $self->_base_offset);
513 $tag = $self->{engine}->create_tag($self, $self->_base_offset, SIG_INDEX, chr(0) x $self->{engine}{index_size});
517 while ($tag->{signature} ne SIG_BLIST) {
518 my $num = ord(substr($md5, $ch, 1));
520 my $ref_loc = $tag->{offset} + ($num * $self->{engine}{long_size});
521 my $new_tag = $self->{engine}->index_lookup($self, $tag, $num);
524 seek($fh, $ref_loc + $self->_root->{file_offset}, SEEK_SET);
525 print( $fh pack($self->{engine}{long_pack}, $self->_root->{end}) );
527 $tag = $self->{engine}->create_tag(
528 $self, $self->_root->{end},
530 chr(0) x $self->{engine}{bucket_list_size},
533 $tag->{ref_loc} = $ref_loc;
541 $tag->{ref_loc} = $ref_loc;
548 # Add key/value to bucket list
550 my $result = $self->{engine}->add_bucket( $self, $tag, $md5, $key, $value );
559 # Fetch single value or element given plain key or array index
561 my $self = shift->_get_self;
564 my $md5 = $self->{engine}{digest}->($key);
567 # Request shared lock for reading
569 $self->lock( LOCK_SH );
571 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
578 # Get value from bucket list
580 my $result = $self->{engine}->get_bucket_value( $self, $tag, $md5 );
584 #XXX What is ref() checking here?
585 #YYY Filters only apply on scalar values, so the ref check is making
586 #YYY sure the fetched bucket is a scalar, not a child hash or array.
587 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
588 ? $self->_root->{filter_fetch_value}->($result)
594 # Delete single key/value pair or element given plain key or array index
596 my $self = $_[0]->_get_self;
599 my $md5 = $self->{engine}{digest}->($key);
602 # Request exclusive lock for writing
604 $self->lock( LOCK_EX );
606 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
615 my $value = $self->{engine}->get_bucket_value($self, $tag, $md5 );
616 if ($value && !ref($value) && $self->_root->{filter_fetch_value}) {
617 $value = $self->_root->{filter_fetch_value}->($value);
620 my $result = $self->{engine}->delete_bucket( $self, $tag, $md5 );
623 # If this object is an array and the key deleted was on the end of the stack,
624 # decrement the length variable.
634 # Check if a single key or element exists given plain key or array index
636 my $self = $_[0]->_get_self;
639 my $md5 = $self->{engine}{digest}->($key);
642 # Request shared lock for reading
644 $self->lock( LOCK_SH );
646 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
649 # For some reason, the built-in exists() function returns '' for false
657 # Check if bucket exists and return 1 or ''
659 my $result = $self->{engine}->bucket_exists( $self, $tag, $md5 ) || '';
668 # Clear all keys from hash, or all elements from array.
670 my $self = $_[0]->_get_self;
673 # Request exclusive lock for writing
675 $self->lock( LOCK_EX );
679 seek($fh, $self->_base_offset + $self->_root->{file_offset}, SEEK_SET);
685 $self->{engine}->create_tag($self, $self->_base_offset, $self->_type, chr(0) x $self->{engine}{index_size});
693 # Public method aliases
695 sub put { (shift)->STORE( @_ ) }
696 sub store { (shift)->STORE( @_ ) }
697 sub get { (shift)->FETCH( @_ ) }
698 sub fetch { (shift)->FETCH( @_ ) }
699 sub delete { (shift)->DELETE( @_ ) }
700 sub exists { (shift)->EXISTS( @_ ) }
701 sub clear { (shift)->CLEAR( @_ ) }
703 package DBM::Deep::_::Root;
717 filter_store_key => undef,
718 filter_store_value => undef,
719 filter_fetch_key => undef,
720 filter_fetch_value => undef,
725 if ( $self->{fh} && !$self->{file_offset} ) {
726 $self->{file_offset} = tell( $self->{fh} );
736 close $self->{fh} if $self->{fh};
747 DBM::Deep - A pure perl multi-level hash/array DBM
752 my $db = DBM::Deep->new( "foo.db" );
754 $db->{key} = 'value'; # tie() style
757 $db->put('key' => 'value'); # OO style
758 print $db->get('key');
760 # true multi-level support
761 $db->{my_complex} = [
762 'hello', { perl => 'rules' },
768 A unique flat-file database module, written in pure perl. True
769 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
770 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
771 handle millions of keys and unlimited hash levels without significant
772 slow-down. Written from the ground-up in pure perl -- this is NOT a
773 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
774 Mac OS X and Windows.
778 Hopefully you are using Perl's excellent CPAN module, which will download
779 and install the module for you. If not, get the tarball, and run these
791 Construction can be done OO-style (which is the recommended way), or using
792 Perl's tie() function. Both are examined here.
794 =head2 OO CONSTRUCTION
796 The recommended way to construct a DBM::Deep object is to use the new()
797 method, which gets you a blessed, tied hash or array reference.
799 my $db = DBM::Deep->new( "foo.db" );
801 This opens a new database handle, mapped to the file "foo.db". If this
802 file does not exist, it will automatically be created. DB files are
803 opened in "r+" (read/write) mode, and the type of object returned is a
804 hash, unless otherwise specified (see L<OPTIONS> below).
806 You can pass a number of options to the constructor to specify things like
807 locking, autoflush, etc. This is done by passing an inline hash:
809 my $db = DBM::Deep->new(
815 Notice that the filename is now specified I<inside> the hash with
816 the "file" parameter, as opposed to being the sole argument to the
817 constructor. This is required if any options are specified.
818 See L<OPTIONS> below for the complete list.
822 You can also start with an array instead of a hash. For this, you must
823 specify the C<type> parameter:
825 my $db = DBM::Deep->new(
827 type => DBM::Deep->TYPE_ARRAY
830 B<Note:> Specifing the C<type> parameter only takes effect when beginning
831 a new DB file. If you create a DBM::Deep object with an existing file, the
832 C<type> will be loaded from the file header, and an error will be thrown if
833 the wrong type is passed in.
835 =head2 TIE CONSTRUCTION
837 Alternately, you can create a DBM::Deep handle by using Perl's built-in
838 tie() function. The object returned from tie() can be used to call methods,
839 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
840 file (as expected with most tie'd objects).
843 my $db = tie %hash, "DBM::Deep", "foo.db";
846 my $db = tie @array, "DBM::Deep", "bar.db";
848 As with the OO constructor, you can replace the DB filename parameter with
849 a hash containing one or more options (see L<OPTIONS> just below for the
852 tie %hash, "DBM::Deep", {
860 There are a number of options that can be passed in when constructing your
861 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
867 Filename of the DB file to link the handle to. You can pass a full absolute
868 filesystem path, partial path, or a plain filename if the file is in the
869 current working directory. This is a required parameter (though q.v. fh).
873 If you want, you can pass in the fh instead of the file. This is most useful for doing
876 my $db = DBM::Deep->new( { fh => \*DATA } );
878 You are responsible for making sure that the fh has been opened appropriately for your
879 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
880 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
881 needs to read from the fh.
885 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
886 not need to set this. However, it's there if you want it.
888 If you pass in fh and do not set this, it will be set appropriately.
892 This parameter specifies what type of object to create, a hash or array. Use
893 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
894 This only takes effect when beginning a new file. This is an optional
895 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
899 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
900 function to lock the database in exclusive mode for writes, and shared mode for
901 reads. Pass any true value to enable. This affects the base DB handle I<and
902 any child hashes or arrays> that use the same DB file. This is an optional
903 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
907 Specifies whether autoflush is to be enabled on the underlying filehandle.
908 This obviously slows down write operations, but is required if you may have
909 multiple processes accessing the same DB file (also consider enable I<locking>).
910 Pass any true value to enable. This is an optional parameter, and defaults to 0
915 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
916 restore them when fetched. This is an B<experimental> feature, and does have
917 side-effects. Basically, when hashes are re-blessed into their original
918 classes, they are no longer blessed into the DBM::Deep class! So you won't be
919 able to call any DBM::Deep methods on them. You have been warned.
920 This is an optional parameter, and defaults to 0 (disabled).
924 See L<FILTERS> below.
930 With DBM::Deep you can access your databases using Perl's standard hash/array
931 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
932 treat them as such. DBM::Deep will intercept all reads/writes and direct them
933 to the right place -- the DB file. This has nothing to do with the
934 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
935 using regular hashes and arrays, rather than calling functions like C<get()>
936 and C<put()> (although those work too). It is entirely up to you how to want
937 to access your databases.
941 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
942 or even nested hashes (or arrays) using standard Perl syntax:
944 my $db = DBM::Deep->new( "foo.db" );
946 $db->{mykey} = "myvalue";
948 $db->{myhash}->{subkey} = "subvalue";
950 print $db->{myhash}->{subkey} . "\n";
952 You can even step through hash keys using the normal Perl C<keys()> function:
954 foreach my $key (keys %$db) {
955 print "$key: " . $db->{$key} . "\n";
958 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
959 pushes them onto an array, all before the loop even begins. If you have an
960 extra large hash, this may exhaust Perl's memory. Instead, consider using
961 Perl's C<each()> function, which pulls keys/values one at a time, using very
964 while (my ($key, $value) = each %$db) {
965 print "$key: $value\n";
968 Please note that when using C<each()>, you should always pass a direct
969 hash reference, not a lookup. Meaning, you should B<never> do this:
972 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
974 This causes an infinite loop, because for each iteration, Perl is calling
975 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
976 it effectively keeps returning the first key over and over again. Instead,
977 assign a temporary variable to C<$db->{foo}>, then pass that to each().
981 As with hashes, you can treat any DBM::Deep object like a normal Perl array
982 reference. This includes inserting, removing and manipulating elements,
983 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
984 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
985 or simply be a nested array reference inside a hash. Example:
987 my $db = DBM::Deep->new(
988 file => "foo-array.db",
989 type => DBM::Deep->TYPE_ARRAY
993 push @$db, "bar", "baz";
996 my $last_elem = pop @$db; # baz
997 my $first_elem = shift @$db; # bah
998 my $second_elem = $db->[1]; # bar
1000 my $num_elements = scalar @$db;
1004 In addition to the I<tie()> interface, you can also use a standard OO interface
1005 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1006 array) has its own methods, but both types share the following common methods:
1007 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1011 =item * new() / clone()
1013 These are the constructor and copy-functions.
1015 =item * put() / store()
1017 Stores a new hash key/value pair, or sets an array element value. Takes two
1018 arguments, the hash key or array index, and the new value. The value can be
1019 a scalar, hash ref or array ref. Returns true on success, false on failure.
1021 $db->put("foo", "bar"); # for hashes
1022 $db->put(1, "bar"); # for arrays
1024 =item * get() / fetch()
1026 Fetches the value of a hash key or array element. Takes one argument: the hash
1027 key or array index. Returns a scalar, hash ref or array ref, depending on the
1030 my $value = $db->get("foo"); # for hashes
1031 my $value = $db->get(1); # for arrays
1035 Checks if a hash key or array index exists. Takes one argument: the hash key
1036 or array index. Returns true if it exists, false if not.
1038 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1039 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1043 Deletes one hash key/value pair or array element. Takes one argument: the hash
1044 key or array index. Returns true on success, false if not found. For arrays,
1045 the remaining elements located after the deleted element are NOT moved over.
1046 The deleted element is essentially just undefined, which is exactly how Perl's
1047 internal arrays work. Please note that the space occupied by the deleted
1048 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1049 below for details and workarounds.
1051 $db->delete("foo"); # for hashes
1052 $db->delete(1); # for arrays
1056 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1057 value. Please note that the space occupied by the deleted keys/values or
1058 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1059 details and workarounds.
1061 $db->clear(); # hashes or arrays
1063 =item * lock() / unlock()
1069 Recover lost disk space.
1071 =item * import() / export()
1073 Data going in and out.
1075 =item * set_digest() / set_pack() / set_filter()
1077 q.v. adjusting the interal parameters.
1083 For hashes, DBM::Deep supports all the common methods described above, and the
1084 following additional methods: C<first_key()> and C<next_key()>.
1090 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1091 fetched in an undefined order (which appears random). Takes no arguments,
1092 returns the key as a scalar value.
1094 my $key = $db->first_key();
1098 Returns the "next" key in the hash, given the previous one as the sole argument.
1099 Returns undef if there are no more keys to be fetched.
1101 $key = $db->next_key($key);
1105 Here are some examples of using hashes:
1107 my $db = DBM::Deep->new( "foo.db" );
1109 $db->put("foo", "bar");
1110 print "foo: " . $db->get("foo") . "\n";
1112 $db->put("baz", {}); # new child hash ref
1113 $db->get("baz")->put("buz", "biz");
1114 print "buz: " . $db->get("baz")->get("buz") . "\n";
1116 my $key = $db->first_key();
1118 print "$key: " . $db->get($key) . "\n";
1119 $key = $db->next_key($key);
1122 if ($db->exists("foo")) { $db->delete("foo"); }
1126 For arrays, DBM::Deep supports all the common methods described above, and the
1127 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1128 C<unshift()> and C<splice()>.
1134 Returns the number of elements in the array. Takes no arguments.
1136 my $len = $db->length();
1140 Adds one or more elements onto the end of the array. Accepts scalars, hash
1141 refs or array refs. No return value.
1143 $db->push("foo", "bar", {});
1147 Fetches the last element in the array, and deletes it. Takes no arguments.
1148 Returns undef if array is empty. Returns the element value.
1150 my $elem = $db->pop();
1154 Fetches the first element in the array, deletes it, then shifts all the
1155 remaining elements over to take up the space. Returns the element value. This
1156 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1159 my $elem = $db->shift();
1163 Inserts one or more elements onto the beginning of the array, shifting all
1164 existing elements over to make room. Accepts scalars, hash refs or array refs.
1165 No return value. This method is not recommended with large arrays -- see
1166 <LARGE ARRAYS> below for details.
1168 $db->unshift("foo", "bar", {});
1172 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1173 -f splice> for usage -- it is too complicated to document here. This method is
1174 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1178 Here are some examples of using arrays:
1180 my $db = DBM::Deep->new(
1182 type => DBM::Deep->TYPE_ARRAY
1185 $db->push("bar", "baz");
1186 $db->unshift("foo");
1189 my $len = $db->length();
1190 print "length: $len\n"; # 4
1192 for (my $k=0; $k<$len; $k++) {
1193 print "$k: " . $db->get($k) . "\n";
1196 $db->splice(1, 2, "biz", "baf");
1198 while (my $elem = shift @$db) {
1199 print "shifted: $elem\n";
1204 Enable automatic file locking by passing a true value to the C<locking>
1205 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1207 my $db = DBM::Deep->new(
1212 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1213 mode for writes, and shared mode for reads. This is required if you have
1214 multiple processes accessing the same database file, to avoid file corruption.
1215 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1216 NFS> below for more.
1218 =head2 EXPLICIT LOCKING
1220 You can explicitly lock a database, so it remains locked for multiple
1221 transactions. This is done by calling the C<lock()> method, and passing an
1222 optional lock mode argument (defaults to exclusive mode). This is particularly
1223 useful for things like counters, where the current value needs to be fetched,
1224 then incremented, then stored again.
1227 my $counter = $db->get("counter");
1229 $db->put("counter", $counter);
1238 You can pass C<lock()> an optional argument, which specifies which mode to use
1239 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1240 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1241 same as the constants defined in Perl's C<Fcntl> module.
1243 $db->lock( DBM::Deep->LOCK_SH );
1247 =head1 IMPORTING/EXPORTING
1249 You can import existing complex structures by calling the C<import()> method,
1250 and export an entire database into an in-memory structure using the C<export()>
1251 method. Both are examined here.
1255 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1256 walking the structure and adding keys/elements to the database as you go,
1257 simply pass a reference to the C<import()> method. This recursively adds
1258 everything to an existing DBM::Deep object for you. Here is an example:
1263 array1 => [ "elem0", "elem1", "elem2" ],
1265 subkey1 => "subvalue1",
1266 subkey2 => "subvalue2"
1270 my $db = DBM::Deep->new( "foo.db" );
1271 $db->import( $struct );
1273 print $db->{key1} . "\n"; # prints "value1"
1275 This recursively imports the entire C<$struct> object into C<$db>, including
1276 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1277 keys are merged with the existing ones, replacing if they already exist.
1278 The C<import()> method can be called on any database level (not just the base
1279 level), and works with both hash and array DB types.
1281 B<Note:> Make sure your existing structure has no circular references in it.
1282 These will cause an infinite loop when importing.
1286 Calling the C<export()> method on an existing DBM::Deep object will return
1287 a reference to a new in-memory copy of the database. The export is done
1288 recursively, so all nested hashes/arrays are all exported to standard Perl
1289 objects. Here is an example:
1291 my $db = DBM::Deep->new( "foo.db" );
1293 $db->{key1} = "value1";
1294 $db->{key2} = "value2";
1296 $db->{hash1}->{subkey1} = "subvalue1";
1297 $db->{hash1}->{subkey2} = "subvalue2";
1299 my $struct = $db->export();
1301 print $struct->{key1} . "\n"; # prints "value1"
1303 This makes a complete copy of the database in memory, and returns a reference
1304 to it. The C<export()> method can be called on any database level (not just
1305 the base level), and works with both hash and array DB types. Be careful of
1306 large databases -- you can store a lot more data in a DBM::Deep object than an
1307 in-memory Perl structure.
1309 B<Note:> Make sure your database has no circular references in it.
1310 These will cause an infinite loop when exporting.
1314 DBM::Deep has a number of hooks where you can specify your own Perl function
1315 to perform filtering on incoming or outgoing data. This is a perfect
1316 way to extend the engine, and implement things like real-time compression or
1317 encryption. Filtering applies to the base DB level, and all child hashes /
1318 arrays. Filter hooks can be specified when your DBM::Deep object is first
1319 constructed, or by calling the C<set_filter()> method at any time. There are
1320 four available filter hooks, described below:
1324 =item * filter_store_key
1326 This filter is called whenever a hash key is stored. It
1327 is passed the incoming key, and expected to return a transformed key.
1329 =item * filter_store_value
1331 This filter is called whenever a hash key or array element is stored. It
1332 is passed the incoming value, and expected to return a transformed value.
1334 =item * filter_fetch_key
1336 This filter is called whenever a hash key is fetched (i.e. via
1337 C<first_key()> or C<next_key()>). It is passed the transformed key,
1338 and expected to return the plain key.
1340 =item * filter_fetch_value
1342 This filter is called whenever a hash key or array element is fetched.
1343 It is passed the transformed value, and expected to return the plain value.
1347 Here are the two ways to setup a filter hook:
1349 my $db = DBM::Deep->new(
1351 filter_store_value => \&my_filter_store,
1352 filter_fetch_value => \&my_filter_fetch
1357 $db->set_filter( "filter_store_value", \&my_filter_store );
1358 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1360 Your filter function will be called only when dealing with SCALAR keys or
1361 values. When nested hashes and arrays are being stored/fetched, filtering
1362 is bypassed. Filters are called as static functions, passed a single SCALAR
1363 argument, and expected to return a single SCALAR value. If you want to
1364 remove a filter, set the function reference to C<undef>:
1366 $db->set_filter( "filter_store_value", undef );
1368 =head2 REAL-TIME ENCRYPTION EXAMPLE
1370 Here is a working example that uses the I<Crypt::Blowfish> module to
1371 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1372 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1373 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1376 use Crypt::Blowfish;
1379 my $cipher = Crypt::CBC->new({
1380 'key' => 'my secret key',
1381 'cipher' => 'Blowfish',
1383 'regenerate_key' => 0,
1384 'padding' => 'space',
1388 my $db = DBM::Deep->new(
1389 file => "foo-encrypt.db",
1390 filter_store_key => \&my_encrypt,
1391 filter_store_value => \&my_encrypt,
1392 filter_fetch_key => \&my_decrypt,
1393 filter_fetch_value => \&my_decrypt,
1396 $db->{key1} = "value1";
1397 $db->{key2} = "value2";
1398 print "key1: " . $db->{key1} . "\n";
1399 print "key2: " . $db->{key2} . "\n";
1405 return $cipher->encrypt( $_[0] );
1408 return $cipher->decrypt( $_[0] );
1411 =head2 REAL-TIME COMPRESSION EXAMPLE
1413 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1414 compression / decompression of keys & values with DBM::Deep Filters.
1415 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1416 more on I<Compress::Zlib>.
1421 my $db = DBM::Deep->new(
1422 file => "foo-compress.db",
1423 filter_store_key => \&my_compress,
1424 filter_store_value => \&my_compress,
1425 filter_fetch_key => \&my_decompress,
1426 filter_fetch_value => \&my_decompress,
1429 $db->{key1} = "value1";
1430 $db->{key2} = "value2";
1431 print "key1: " . $db->{key1} . "\n";
1432 print "key2: " . $db->{key2} . "\n";
1438 return Compress::Zlib::memGzip( $_[0] ) ;
1441 return Compress::Zlib::memGunzip( $_[0] ) ;
1444 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1445 actually numerical index numbers, and are not filtered.
1447 =head1 ERROR HANDLING
1449 Most DBM::Deep methods return a true value for success, and call die() on
1450 failure. You can wrap calls in an eval block to catch the die.
1452 my $db = DBM::Deep->new( "foo.db" ); # create hash
1453 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1455 print $@; # prints error message
1457 =head1 LARGEFILE SUPPORT
1459 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1460 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1461 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1462 by calling the static C<set_pack()> method before you do anything else.
1464 DBM::Deep::set_pack(8, 'Q');
1466 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1467 instead of 32-bit longs. After setting these values your DB files have a
1468 theoretical maximum size of 16 XB (exabytes).
1470 B<Note:> Changing these values will B<NOT> work for existing database files.
1471 Only change this for new files, and make sure it stays set consistently
1472 throughout the file's life. If you do set these values, you can no longer
1473 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
1474 back to 32-bit mode.
1476 B<Note:> I have not personally tested files > 2 GB -- all my systems have
1477 only a 32-bit Perl. However, I have received user reports that this does
1480 =head1 LOW-LEVEL ACCESS
1482 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1483 you can call the C<_fh()> method, which returns the handle:
1485 my $fh = $db->_fh();
1487 This method can be called on the root level of the datbase, or any child
1488 hashes or arrays. All levels share a I<root> structure, which contains things
1489 like the filehandle, a reference counter, and all the options specified
1490 when you created the object. You can get access to this root structure by
1491 calling the C<root()> method.
1493 my $root = $db->_root();
1495 This is useful for changing options after the object has already been created,
1496 such as enabling/disabling locking. You can also store your own temporary user
1497 data in this structure (be wary of name collision), which is then accessible from
1498 any child hash or array.
1500 =head1 CUSTOM DIGEST ALGORITHM
1502 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1503 keys. However you can override this, and use another algorithm (such as SHA-256)
1504 or even write your own. But please note that DBM::Deep currently expects zero
1505 collisions, so your algorithm has to be I<perfect>, so to speak.
1506 Collision detection may be introduced in a later version.
1510 You can specify a custom digest algorithm by calling the static C<set_digest()>
1511 function, passing a reference to a subroutine, and the length of the algorithm's
1512 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
1513 objects. Here is a working example that uses a 256-bit hash from the
1514 I<Digest::SHA256> module. Please see
1515 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
1520 my $context = Digest::SHA256::new(256);
1522 DBM::Deep::set_digest( \&my_digest, 32 );
1524 my $db = DBM::Deep->new( "foo-sha.db" );
1526 $db->{key1} = "value1";
1527 $db->{key2} = "value2";
1528 print "key1: " . $db->{key1} . "\n";
1529 print "key2: " . $db->{key2} . "\n";
1535 return substr( $context->hash($_[0]), 0, 32 );
1538 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1539 of bytes you specify in the C<set_digest()> function (in this case 32).
1541 =head1 CIRCULAR REFERENCES
1543 DBM::Deep has B<experimental> support for circular references. Meaning you
1544 can have a nested hash key or array element that points to a parent object.
1545 This relationship is stored in the DB file, and is preserved between sessions.
1548 my $db = DBM::Deep->new( "foo.db" );
1551 $db->{circle} = $db; # ref to self
1553 print $db->{foo} . "\n"; # prints "foo"
1554 print $db->{circle}->{foo} . "\n"; # prints "foo" again
1556 One catch is, passing the object to a function that recursively walks the
1557 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1558 C<export()> methods) will result in an infinite loop. The other catch is,
1559 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
1560 or C<next_key()> methods), you will get the I<target object's key>, not the
1561 ref's key. This gets even more interesting with the above example, where
1562 the I<circle> key points to the base DB object, which technically doesn't
1563 have a key. So I made DBM::Deep return "[base]" as the key name in that
1566 =head1 CAVEATS / ISSUES / BUGS
1568 This section describes all the known issues with DBM::Deep. It you have found
1569 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1571 =head2 UNUSED SPACE RECOVERY
1573 One major caveat with DBM::Deep is that space occupied by existing keys and
1574 values is not recovered when they are deleted. Meaning if you keep deleting
1575 and adding new keys, your file will continuously grow. I am working on this,
1576 but in the meantime you can call the built-in C<optimize()> method from time to
1577 time (perhaps in a crontab or something) to recover all your unused space.
1579 $db->optimize(); # returns true on success
1581 This rebuilds the ENTIRE database into a new file, then moves it on top of
1582 the original. The new file will have no unused space, thus it will take up as
1583 little disk space as possible. Please note that this operation can take
1584 a long time for large files, and you need enough disk space to temporarily hold
1585 2 copies of your DB file. The temporary file is created in the same directory
1586 as the original, named with a ".tmp" extension, and is deleted when the
1587 operation completes. Oh, and if locking is enabled, the DB is automatically
1588 locked for the entire duration of the copy.
1590 B<WARNING:> Only call optimize() on the top-level node of the database, and
1591 make sure there are no child references lying around. DBM::Deep keeps a reference
1592 counter, and if it is greater than 1, optimize() will abort and return undef.
1594 =head2 AUTOVIVIFICATION
1596 Unfortunately, autovivification doesn't work with tied hashes. This appears to
1597 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
1598 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
1599 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
1600 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
1603 $db->{foo}->{bar} = "hello";
1605 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
1606 being an empty hash. Try this instead, which works fine:
1608 $db->{foo} = { bar => "hello" };
1610 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
1611 the execution path, and Perl indeed passes an empty hash to the STORE() method.
1612 Probably a bug in Perl.
1614 =head2 FILE CORRUPTION
1616 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1617 for a 32-bit signature when opened, but other corruption in files can cause
1618 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1619 stuck in an infinite loop depending on the level of corruption. File write
1620 operations are not checked for failure (for speed), so if you happen to run
1621 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1622 be addressed in a later version of DBM::Deep.
1626 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1627 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1628 about setting up your NFS server with a locking daemon, then using lockf() to
1629 lock your files, but your mileage may vary there as well. From what I
1630 understand, there is no real way to do it. However, if you need access to the
1631 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1632 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1634 =head2 COPYING OBJECTS
1636 Beware of copying tied objects in Perl. Very strange things can happen.
1637 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1638 returns a new, blessed, tied hash or array to the same level in the DB.
1640 my $copy = $db->clone();
1642 B<Note>: Since clone() here is cloning the object, not the database location, any
1643 modifications to either $db or $copy will be visible in both.
1647 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1648 These functions cause every element in the array to move, which can be murder
1649 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1650 a different location. This will be addressed in the forthcoming version 1.00.
1652 =head2 WRITEONLY FILES
1654 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1655 writeonly mode. STORE will verify that the filehandle is writable. However, there
1656 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1657 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1661 This section discusses DBM::Deep's speed and memory usage.
1665 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1666 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1667 multi-level hash/array support, and cross-platform FTPable files. Even so,
1668 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1669 with huge databases. Here is some test data:
1671 Adding 1,000,000 keys to new DB file...
1673 At 100 keys, avg. speed is 2,703 keys/sec
1674 At 200 keys, avg. speed is 2,642 keys/sec
1675 At 300 keys, avg. speed is 2,598 keys/sec
1676 At 400 keys, avg. speed is 2,578 keys/sec
1677 At 500 keys, avg. speed is 2,722 keys/sec
1678 At 600 keys, avg. speed is 2,628 keys/sec
1679 At 700 keys, avg. speed is 2,700 keys/sec
1680 At 800 keys, avg. speed is 2,607 keys/sec
1681 At 900 keys, avg. speed is 2,190 keys/sec
1682 At 1,000 keys, avg. speed is 2,570 keys/sec
1683 At 2,000 keys, avg. speed is 2,417 keys/sec
1684 At 3,000 keys, avg. speed is 1,982 keys/sec
1685 At 4,000 keys, avg. speed is 1,568 keys/sec
1686 At 5,000 keys, avg. speed is 1,533 keys/sec
1687 At 6,000 keys, avg. speed is 1,787 keys/sec
1688 At 7,000 keys, avg. speed is 1,977 keys/sec
1689 At 8,000 keys, avg. speed is 2,028 keys/sec
1690 At 9,000 keys, avg. speed is 2,077 keys/sec
1691 At 10,000 keys, avg. speed is 2,031 keys/sec
1692 At 20,000 keys, avg. speed is 1,970 keys/sec
1693 At 30,000 keys, avg. speed is 2,050 keys/sec
1694 At 40,000 keys, avg. speed is 2,073 keys/sec
1695 At 50,000 keys, avg. speed is 1,973 keys/sec
1696 At 60,000 keys, avg. speed is 1,914 keys/sec
1697 At 70,000 keys, avg. speed is 2,091 keys/sec
1698 At 80,000 keys, avg. speed is 2,103 keys/sec
1699 At 90,000 keys, avg. speed is 1,886 keys/sec
1700 At 100,000 keys, avg. speed is 1,970 keys/sec
1701 At 200,000 keys, avg. speed is 2,053 keys/sec
1702 At 300,000 keys, avg. speed is 1,697 keys/sec
1703 At 400,000 keys, avg. speed is 1,838 keys/sec
1704 At 500,000 keys, avg. speed is 1,941 keys/sec
1705 At 600,000 keys, avg. speed is 1,930 keys/sec
1706 At 700,000 keys, avg. speed is 1,735 keys/sec
1707 At 800,000 keys, avg. speed is 1,795 keys/sec
1708 At 900,000 keys, avg. speed is 1,221 keys/sec
1709 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1711 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1712 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1713 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1714 Run time was 12 min 3 sec.
1718 One of the great things about DBM::Deep is that it uses very little memory.
1719 Even with huge databases (1,000,000+ keys) you will not see much increased
1720 memory on your process. DBM::Deep relies solely on the filesystem for storing
1721 and fetching data. Here is output from I</usr/bin/top> before even opening a
1724 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1725 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1727 Basically the process is taking 2,716K of memory. And here is the same
1728 process after storing and fetching 1,000,000 keys:
1730 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1731 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1733 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1734 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1736 =head1 DB FILE FORMAT
1738 In case you were interested in the underlying DB file format, it is documented
1739 here in this section. You don't need to know this to use the module, it's just
1740 included for reference.
1744 DBM::Deep files always start with a 32-bit signature to identify the file type.
1745 This is at offset 0. The signature is "DPDB" in network byte order. This is
1746 checked for when the file is opened and an error will be thrown if it's not found.
1750 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
1751 has a standard header containing the type of data, the length of data, and then
1752 the data itself. The type is a single character (1 byte), the length is a
1753 32-bit unsigned long in network byte order, and the data is, well, the data.
1754 Here is how it unfolds:
1758 Immediately after the 32-bit file signature is the I<Master Index> record.
1759 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
1760 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
1761 depending on how the DBM::Deep object was constructed.
1763 The index works by looking at a I<MD5 Hash> of the hash key (or array index
1764 number). The first 8-bit char of the MD5 signature is the offset into the
1765 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
1766 index element is a file offset of the next tag for the key/element in question,
1767 which is usually a I<Bucket List> tag (see below).
1769 The next tag I<could> be another index, depending on how many keys/elements
1770 exist. See L<RE-INDEXING> below for details.
1774 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
1775 file offsets to where the actual data is stored. It starts with a standard
1776 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
1777 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
1778 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
1779 When the list fills up, a I<Re-Index> operation is performed (See
1780 L<RE-INDEXING> below).
1784 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
1785 index/value pair (in array mode). It starts with a standard tag header with
1786 type I<D> for scalar data (string, binary, etc.), or it could be a nested
1787 hash (type I<H>) or array (type I<A>). The value comes just after the tag
1788 header. The size reported in the tag header is only for the value, but then,
1789 just after the value is another size (32-bit unsigned long) and then the plain
1790 key itself. Since the value is likely to be fetched more often than the plain
1791 key, I figured it would be I<slightly> faster to store the value first.
1793 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
1794 record for the nested structure, where the process begins all over again.
1798 After a I<Bucket List> grows to 16 records, its allocated space in the file is
1799 exhausted. Then, when another key/element comes in, the list is converted to a
1800 new index record. However, this index will look at the next char in the MD5
1801 hash, and arrange new Bucket List pointers accordingly. This process is called
1802 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
1803 17 (16 + new one) keys/elements are removed from the old Bucket List and
1804 inserted into the new index. Several new Bucket Lists are created in the
1805 process, as a new MD5 char from the key is being examined (it is unlikely that
1806 the keys will all share the same next char of their MD5s).
1808 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
1809 when the Bucket Lists will turn into indexes, but the first round tends to
1810 happen right around 4,000 keys. You will see a I<slight> decrease in
1811 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
1812 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
1813 right around 900,000 keys. This process can continue nearly indefinitely --
1814 right up until the point the I<MD5> signatures start colliding with each other,
1815 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
1816 getting struck by lightning while you are walking to cash in your tickets.
1817 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
1818 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
1819 this is 340 unodecillion, but don't quote me).
1823 When a new key/element is stored, the key (or index number) is first run through
1824 I<Digest::MD5> to get a 128-bit signature (example, in hex:
1825 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
1826 for the first char of the signature (in this case I<b0>). If it does not exist,
1827 a new I<Bucket List> is created for our key (and the next 15 future keys that
1828 happen to also have I<b> as their first MD5 char). The entire MD5 is written
1829 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
1830 this point, unless we are replacing an existing I<Bucket>), where the actual
1831 data will be stored.
1835 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
1836 (or index number), then walking along the indexes. If there are enough
1837 keys/elements in this DB level, there might be nested indexes, each linked to
1838 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
1839 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
1840 question. If we found a match, the I<Bucket> tag is loaded, where the value and
1841 plain key are stored.
1843 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
1844 methods. In this process the indexes are walked systematically, and each key
1845 fetched in increasing MD5 order (which is why it appears random). Once the
1846 I<Bucket> is found, the value is skipped and the plain key returned instead.
1847 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
1848 alphabetically sorted. This only happens on an index-level -- as soon as the
1849 I<Bucket Lists> are hit, the keys will come out in the order they went in --
1850 so it's pretty much undefined how the keys will come out -- just like Perl's
1853 =head1 CODE COVERAGE
1855 We use B<Devel::Cover> to test the code coverage of our tests, below is the
1856 B<Devel::Cover> report on this module's test suite.
1858 ---------------------------- ------ ------ ------ ------ ------ ------ ------
1859 File stmt bran cond sub pod time total
1860 ---------------------------- ------ ------ ------ ------ ------ ------ ------
1861 blib/lib/DBM/Deep.pm 95.1 81.6 70.3 100.0 100.0 33.4 91.0
1862 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 n/a 27.8 98.0
1863 blib/lib/DBM/Deep/Engine.pm 97.8 85.6 75.0 100.0 0.0 25.8 90.8
1864 blib/lib/DBM/Deep/Hash.pm 100.0 87.5 100.0 100.0 n/a 13.0 97.2
1865 Total 97.5 85.4 76.6 100.0 46.9 100.0 92.5
1866 ---------------------------- ------ ------ ------ ------ ------ ------ ------
1868 =head1 MORE INFORMATION
1870 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
1871 or send email to L<DBM-Deep@googlegroups.com>.
1875 Joseph Huckaby, L<jhuckaby@cpan.org>
1877 Rob Kinyon, L<rkinyon@cpan.org>
1879 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
1883 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
1884 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
1888 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
1889 This is free software, you may use it and distribute it under the
1890 same terms as Perl itself.