7 # Multi-level database module for storing hash trees, arrays and simple
8 # key/value pairs into FTP-able, cross-platform binary database files.
10 # Type `perldoc DBM::Deep` for complete documentation.
14 # tie %db, 'DBM::Deep', 'my_database.db'; # standard tie() method
16 # my $db = new DBM::Deep( 'my_database.db' ); # preferred OO method
18 # $db->{my_scalar} = 'hello world';
19 # $db->{my_hash} = { larry => 'genius', hashes => 'fast' };
20 # $db->{my_array} = [ 1, 2, 3, time() ];
21 # $db->{my_complex} = [ 'hello', { perl => 'rules' }, 42, 99 ];
22 # push @{$db->{my_array}}, 'another value';
23 # my @key_list = keys %{$db->{my_hash}};
24 # print "This module " . $db->{my_complex}->[1]->{perl} . "!\n";
27 # (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
28 # This program is free software; you can redistribute it and/or
29 # modify it under the same terms as Perl itself.
34 use Fcntl qw( :DEFAULT :flock :seek );
38 use DBM::Deep::Engine;
40 use vars qw( $VERSION );
41 $VERSION = q(0.99_01);
44 # Setup constants for users to pass to new()
46 sub TYPE_HASH () { DBM::Deep::Engine::SIG_HASH }
47 sub TYPE_ARRAY () { DBM::Deep::Engine::SIG_ARRAY }
48 sub TYPE_SCALAR () { DBM::Deep::Engine::SIG_SCALAR }
56 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
61 unless ( eval { local $SIG{'__DIE__'}; %{$_[0]} || 1 } ) {
62 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
67 $args = { file => shift };
75 # Class constructor method for Perl OO interface.
76 # Calls tie() and returns blessed reference to tied hash or array,
77 # providing a hybrid OO/tie interface.
80 my $args = $class->_get_args( @_ );
83 # Check if we want a tied hash or array.
86 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
87 $class = 'DBM::Deep::Array';
88 require DBM::Deep::Array;
89 tie @$self, $class, %$args;
92 $class = 'DBM::Deep::Hash';
93 require DBM::Deep::Hash;
94 tie %$self, $class, %$args;
97 return bless $self, $class;
102 # Setup $self and bless into this class.
107 # These are the defaults to be optionally overridden below
110 engine => DBM::Deep::Engine->new,
112 $self->{base_offset} = length( $self->{engine}->SIG_FILE );
114 foreach my $param ( keys %$self ) {
115 next unless exists $args->{$param};
116 $self->{$param} = delete $args->{$param}
119 # locking implicitly enables autoflush
120 if ($args->{locking}) { $args->{autoflush} = 1; }
122 $self->{root} = exists $args->{root}
124 : DBM::Deep::_::Root->new( $args );
126 $self->{engine}->setup_fh( $self );
133 require DBM::Deep::Hash;
134 return DBM::Deep::Hash->TIEHASH( @_ );
139 require DBM::Deep::Array;
140 return DBM::Deep::Array->TIEARRAY( @_ );
143 #XXX Unneeded now ...
149 # If db locking is set, flock() the db file. If called multiple
150 # times before unlock(), then the same number of unlocks() must
151 # be called before the lock is released.
153 my $self = shift->_get_self;
155 $type = LOCK_EX unless defined $type;
157 if (!defined($self->_fh)) { return; }
159 if ($self->_root->{locking}) {
160 if (!$self->_root->{locked}) {
161 flock($self->_fh, $type);
163 # refresh end counter in case file has changed size
164 my @stats = stat($self->_root->{file});
165 $self->_root->{end} = $stats[7];
167 # double-check file inode, in case another process
168 # has optimize()d our file while we were waiting.
169 if ($stats[1] != $self->_root->{inode}) {
170 $self->{engine}->close_fh( $self );
171 $self->{engine}->setup_fh( $self );
172 flock($self->_fh, $type); # re-lock
174 # This may not be necessary after re-opening
175 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
178 $self->_root->{locked}++;
188 # If db locking is set, unlock the db file. See note in lock()
189 # regarding calling lock() multiple times.
191 my $self = shift->_get_self;
193 if (!defined($self->_fh)) { return; }
195 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
196 $self->_root->{locked}--;
197 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
206 my $self = shift->_get_self;
207 my ($spot, $value) = @_;
212 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
213 my $type = $value->_type;
214 ${$spot} = $type eq TYPE_HASH ? {} : [];
215 $value->_copy_node( ${$spot} );
218 my $r = Scalar::Util::reftype( $value );
219 my $c = Scalar::Util::blessed( $value );
220 if ( $r eq 'ARRAY' ) {
221 ${$spot} = [ @{$value} ];
224 ${$spot} = { %{$value} };
226 ${$spot} = bless ${$spot}, $c
235 # Copy single level of keys or elements to new DB handle.
236 # Recurse for nested structures
238 my $self = shift->_get_self;
241 if ($self->_type eq TYPE_HASH) {
242 my $key = $self->first_key();
244 my $value = $self->get($key);
245 $self->_copy_value( \$db_temp->{$key}, $value );
246 $key = $self->next_key($key);
250 my $length = $self->length();
251 for (my $index = 0; $index < $length; $index++) {
252 my $value = $self->get($index);
253 $self->_copy_value( \$db_temp->[$index], $value );
262 # Recursively export into standard Perl hashes and arrays.
264 my $self = shift->_get_self;
267 if ($self->_type eq TYPE_HASH) { $temp = {}; }
268 elsif ($self->_type eq TYPE_ARRAY) { $temp = []; }
271 $self->_copy_node( $temp );
279 # Recursively import Perl hash/array structure
281 #XXX This use of ref() seems to be ok
282 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
284 my $self = shift->_get_self;
287 #XXX This use of ref() seems to be ok
290 # struct is not a reference, so just import based on our type
294 if ($self->_type eq TYPE_HASH) { $struct = {@_}; }
295 elsif ($self->_type eq TYPE_ARRAY) { $struct = [@_]; }
298 my $r = Scalar::Util::reftype($struct) || '';
299 if ($r eq "HASH" && $self->_type eq TYPE_HASH) {
300 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
302 elsif ($r eq "ARRAY" && $self->_type eq TYPE_ARRAY) {
303 $self->push( @$struct );
306 $self->_throw_error("Cannot import: type mismatch");
314 # Rebuild entire database into new file, then move
315 # it back on top of original.
317 my $self = shift->_get_self;
319 #XXX Need to create a new test for this
320 # if ($self->_root->{links} > 1) {
321 # $self->_throw_error("Cannot optimize: reference count is greater than 1");
324 my $db_temp = DBM::Deep->new(
325 file => $self->_root->{file} . '.tmp',
329 $self->_throw_error("Cannot optimize: failed to open temp file: $!");
333 $self->_copy_node( $db_temp );
337 # Attempt to copy user, group and permissions over to new file
339 my @stats = stat($self->_fh);
340 my $perms = $stats[2] & 07777;
343 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
344 chmod( $perms, $self->_root->{file} . '.tmp' );
346 # q.v. perlport for more information on this variable
347 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
349 # Potential race condition when optmizing on Win32 with locking.
350 # The Windows filesystem requires that the filehandle be closed
351 # before it is overwritten with rename(). This could be redone
355 $self->{engine}->close_fh( $self );
358 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
359 unlink $self->_root->{file} . '.tmp';
361 $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
365 $self->{engine}->close_fh( $self );
366 $self->{engine}->setup_fh( $self );
373 # Make copy of object and return
375 my $self = shift->_get_self;
377 return DBM::Deep->new(
378 type => $self->_type,
379 base_offset => $self->_base_offset,
385 my %is_legal_filter = map {
388 store_key store_value
389 fetch_key fetch_value
394 # Setup filter function for storing or fetching the key or value
396 my $self = shift->_get_self;
400 if ( $is_legal_filter{$type} ) {
401 $self->_root->{"filter_$type"} = $func;
415 # Get access to the root structure
417 my $self = $_[0]->_get_self;
418 return $self->{root};
423 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
425 my $self = $_[0]->_get_self;
426 return $self->{type};
431 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
433 my $self = $_[0]->_get_self;
434 return $self->{base_offset};
439 # Get access to the raw fh
441 my $self = $_[0]->_get_self;
442 return $self->_root->{fh};
450 die "DBM::Deep: $_[1]\n";
455 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
460 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
465 # Store single hash key/value or array element in database.
467 my $self = shift->_get_self;
468 my ($key, $value) = @_;
470 unless ( _is_writable( $self->_fh ) ) {
471 $self->_throw_error( 'Cannot write to a readonly filehandle' );
475 # Request exclusive lock for writing
477 $self->lock( LOCK_EX );
479 my $md5 = $self->{engine}{digest}->($key);
481 my $tag = $self->{engine}->find_bucket_list( $self, $md5, { create => 1 } );
483 # User may be storing a hash, in which case we do not want it run
484 # through the filtering system
485 if ( !ref($value) && $self->_root->{filter_store_value} ) {
486 $value = $self->_root->{filter_store_value}->( $value );
490 # Add key/value to bucket list
492 my $result = $self->{engine}->add_bucket( $self, $tag, $md5, $key, $value );
501 # Fetch single value or element given plain key or array index
503 my $self = shift->_get_self;
506 my $md5 = $self->{engine}{digest}->($key);
509 # Request shared lock for reading
511 $self->lock( LOCK_SH );
513 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
520 # Get value from bucket list
522 my $result = $self->{engine}->get_bucket_value( $self, $tag, $md5 );
526 # Filters only apply to scalar values, so the ref check is making
527 # sure the fetched bucket is a scalar, not a child hash or array.
528 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
529 ? $self->_root->{filter_fetch_value}->($result)
535 # Delete single key/value pair or element given plain key or array index
537 my $self = $_[0]->_get_self;
540 unless ( _is_writable( $self->_fh ) ) {
541 $self->_throw_error( 'Cannot write to a readonly filehandle' );
545 # Request exclusive lock for writing
547 $self->lock( LOCK_EX );
549 my $md5 = $self->{engine}{digest}->($key);
551 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
560 my $value = $self->{engine}->get_bucket_value($self, $tag, $md5 );
562 if (defined $value && !ref($value) && $self->_root->{filter_fetch_value}) {
563 $value = $self->_root->{filter_fetch_value}->($value);
566 my $result = $self->{engine}->delete_bucket( $self, $tag, $md5 );
569 # If this object is an array and the key deleted was on the end of the stack,
570 # decrement the length variable.
580 # Check if a single key or element exists given plain key or array index
582 my $self = $_[0]->_get_self;
585 my $md5 = $self->{engine}{digest}->($key);
588 # Request shared lock for reading
590 $self->lock( LOCK_SH );
592 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
597 # For some reason, the built-in exists() function returns '' for false
603 # Check if bucket exists and return 1 or ''
605 my $result = $self->{engine}->bucket_exists( $self, $tag, $md5 ) || '';
614 # Clear all keys from hash, or all elements from array.
616 my $self = $_[0]->_get_self;
618 unless ( _is_writable( $self->_fh ) ) {
619 $self->_throw_error( 'Cannot write to a readonly filehandle' );
623 # Request exclusive lock for writing
625 $self->lock( LOCK_EX );
629 seek($fh, $self->_base_offset + $self->_root->{file_offset}, SEEK_SET);
635 $self->{engine}->create_tag($self, $self->_base_offset, $self->_type, chr(0) x $self->{engine}{index_size});
643 # Public method aliases
645 sub put { (shift)->STORE( @_ ) }
646 sub store { (shift)->STORE( @_ ) }
647 sub get { (shift)->FETCH( @_ ) }
648 sub fetch { (shift)->FETCH( @_ ) }
649 sub delete { (shift)->DELETE( @_ ) }
650 sub exists { (shift)->EXISTS( @_ ) }
651 sub clear { (shift)->CLEAR( @_ ) }
653 package DBM::Deep::_::Root;
662 #XXX It should be this in order to work with the initial create_tag(),
663 #XXX but it's not ... it works out because of the stat() in setup_fh(),
664 #XXX but that's not good.
665 end => 0, #length(DBM::Deep->SIG_FILE),
671 filter_store_key => undef,
672 filter_store_value => undef,
673 filter_fetch_key => undef,
674 filter_fetch_value => undef,
678 if ( $self->{fh} && !$self->{file_offset} ) {
679 $self->{file_offset} = tell( $self->{fh} );
689 close $self->{fh} if $self->{fh};
699 DBM::Deep - A pure perl multi-level hash/array DBM
704 my $db = DBM::Deep->new( "foo.db" );
706 $db->{key} = 'value'; # tie() style
709 $db->put('key' => 'value'); # OO style
710 print $db->get('key');
712 # true multi-level support
713 $db->{my_complex} = [
714 'hello', { perl => 'rules' },
720 A unique flat-file database module, written in pure perl. True
721 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
722 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
723 handle millions of keys and unlimited hash levels without significant
724 slow-down. Written from the ground-up in pure perl -- this is NOT a
725 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
726 Mac OS X and Windows.
728 =head1 VERSION DIFFERENCES
730 B<NOTE>: 0.99_01 and above have significant file format differences from 0.98 and
731 before. While attempts have been made to be backwards compatible, no guarantees.
735 Hopefully you are using Perl's excellent CPAN module, which will download
736 and install the module for you. If not, get the tarball, and run these
748 Construction can be done OO-style (which is the recommended way), or using
749 Perl's tie() function. Both are examined here.
751 =head2 OO CONSTRUCTION
753 The recommended way to construct a DBM::Deep object is to use the new()
754 method, which gets you a blessed, tied hash or array reference.
756 my $db = DBM::Deep->new( "foo.db" );
758 This opens a new database handle, mapped to the file "foo.db". If this
759 file does not exist, it will automatically be created. DB files are
760 opened in "r+" (read/write) mode, and the type of object returned is a
761 hash, unless otherwise specified (see L<OPTIONS> below).
763 You can pass a number of options to the constructor to specify things like
764 locking, autoflush, etc. This is done by passing an inline hash:
766 my $db = DBM::Deep->new(
772 Notice that the filename is now specified I<inside> the hash with
773 the "file" parameter, as opposed to being the sole argument to the
774 constructor. This is required if any options are specified.
775 See L<OPTIONS> below for the complete list.
779 You can also start with an array instead of a hash. For this, you must
780 specify the C<type> parameter:
782 my $db = DBM::Deep->new(
784 type => DBM::Deep->TYPE_ARRAY
787 B<Note:> Specifing the C<type> parameter only takes effect when beginning
788 a new DB file. If you create a DBM::Deep object with an existing file, the
789 C<type> will be loaded from the file header, and an error will be thrown if
790 the wrong type is passed in.
792 =head2 TIE CONSTRUCTION
794 Alternately, you can create a DBM::Deep handle by using Perl's built-in
795 tie() function. The object returned from tie() can be used to call methods,
796 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
797 file (as expected with most tie'd objects).
800 my $db = tie %hash, "DBM::Deep", "foo.db";
803 my $db = tie @array, "DBM::Deep", "bar.db";
805 As with the OO constructor, you can replace the DB filename parameter with
806 a hash containing one or more options (see L<OPTIONS> just below for the
809 tie %hash, "DBM::Deep", {
817 There are a number of options that can be passed in when constructing your
818 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
824 Filename of the DB file to link the handle to. You can pass a full absolute
825 filesystem path, partial path, or a plain filename if the file is in the
826 current working directory. This is a required parameter (though q.v. fh).
830 If you want, you can pass in the fh instead of the file. This is most useful for doing
833 my $db = DBM::Deep->new( { fh => \*DATA } );
835 You are responsible for making sure that the fh has been opened appropriately for your
836 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
837 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
838 needs to read from the fh.
842 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
843 not need to set this. However, it's there if you want it.
845 If you pass in fh and do not set this, it will be set appropriately.
849 This parameter specifies what type of object to create, a hash or array. Use
850 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
851 This only takes effect when beginning a new file. This is an optional
852 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
856 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
857 function to lock the database in exclusive mode for writes, and shared mode for
858 reads. Pass any true value to enable. This affects the base DB handle I<and
859 any child hashes or arrays> that use the same DB file. This is an optional
860 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
864 Specifies whether autoflush is to be enabled on the underlying filehandle.
865 This obviously slows down write operations, but is required if you may have
866 multiple processes accessing the same DB file (also consider enable I<locking>).
867 Pass any true value to enable. This is an optional parameter, and defaults to 0
872 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
873 restore them when fetched. This is an B<experimental> feature, and does have
874 side-effects. Basically, when hashes are re-blessed into their original
875 classes, they are no longer blessed into the DBM::Deep class! So you won't be
876 able to call any DBM::Deep methods on them. You have been warned.
877 This is an optional parameter, and defaults to 0 (disabled).
881 See L<FILTERS> below.
887 With DBM::Deep you can access your databases using Perl's standard hash/array
888 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
889 treat them as such. DBM::Deep will intercept all reads/writes and direct them
890 to the right place -- the DB file. This has nothing to do with the
891 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
892 using regular hashes and arrays, rather than calling functions like C<get()>
893 and C<put()> (although those work too). It is entirely up to you how to want
894 to access your databases.
898 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
899 or even nested hashes (or arrays) using standard Perl syntax:
901 my $db = DBM::Deep->new( "foo.db" );
903 $db->{mykey} = "myvalue";
905 $db->{myhash}->{subkey} = "subvalue";
907 print $db->{myhash}->{subkey} . "\n";
909 You can even step through hash keys using the normal Perl C<keys()> function:
911 foreach my $key (keys %$db) {
912 print "$key: " . $db->{$key} . "\n";
915 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
916 pushes them onto an array, all before the loop even begins. If you have an
917 extra large hash, this may exhaust Perl's memory. Instead, consider using
918 Perl's C<each()> function, which pulls keys/values one at a time, using very
921 while (my ($key, $value) = each %$db) {
922 print "$key: $value\n";
925 Please note that when using C<each()>, you should always pass a direct
926 hash reference, not a lookup. Meaning, you should B<never> do this:
929 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
931 This causes an infinite loop, because for each iteration, Perl is calling
932 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
933 it effectively keeps returning the first key over and over again. Instead,
934 assign a temporary variable to C<$db->{foo}>, then pass that to each().
938 As with hashes, you can treat any DBM::Deep object like a normal Perl array
939 reference. This includes inserting, removing and manipulating elements,
940 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
941 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
942 or simply be a nested array reference inside a hash. Example:
944 my $db = DBM::Deep->new(
945 file => "foo-array.db",
946 type => DBM::Deep->TYPE_ARRAY
950 push @$db, "bar", "baz";
953 my $last_elem = pop @$db; # baz
954 my $first_elem = shift @$db; # bah
955 my $second_elem = $db->[1]; # bar
957 my $num_elements = scalar @$db;
961 In addition to the I<tie()> interface, you can also use a standard OO interface
962 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
963 array) has its own methods, but both types share the following common methods:
964 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
968 =item * new() / clone()
970 These are the constructor and copy-functions.
972 =item * put() / store()
974 Stores a new hash key/value pair, or sets an array element value. Takes two
975 arguments, the hash key or array index, and the new value. The value can be
976 a scalar, hash ref or array ref. Returns true on success, false on failure.
978 $db->put("foo", "bar"); # for hashes
979 $db->put(1, "bar"); # for arrays
981 =item * get() / fetch()
983 Fetches the value of a hash key or array element. Takes one argument: the hash
984 key or array index. Returns a scalar, hash ref or array ref, depending on the
987 my $value = $db->get("foo"); # for hashes
988 my $value = $db->get(1); # for arrays
992 Checks if a hash key or array index exists. Takes one argument: the hash key
993 or array index. Returns true if it exists, false if not.
995 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
996 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1000 Deletes one hash key/value pair or array element. Takes one argument: the hash
1001 key or array index. Returns true on success, false if not found. For arrays,
1002 the remaining elements located after the deleted element are NOT moved over.
1003 The deleted element is essentially just undefined, which is exactly how Perl's
1004 internal arrays work. Please note that the space occupied by the deleted
1005 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1006 below for details and workarounds.
1008 $db->delete("foo"); # for hashes
1009 $db->delete(1); # for arrays
1013 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1014 value. Please note that the space occupied by the deleted keys/values or
1015 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1016 details and workarounds.
1018 $db->clear(); # hashes or arrays
1020 =item * lock() / unlock()
1026 Recover lost disk space.
1028 =item * import() / export()
1030 Data going in and out.
1032 =item * set_digest() / set_pack() / set_filter()
1034 q.v. adjusting the interal parameters.
1040 For hashes, DBM::Deep supports all the common methods described above, and the
1041 following additional methods: C<first_key()> and C<next_key()>.
1047 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1048 fetched in an undefined order (which appears random). Takes no arguments,
1049 returns the key as a scalar value.
1051 my $key = $db->first_key();
1055 Returns the "next" key in the hash, given the previous one as the sole argument.
1056 Returns undef if there are no more keys to be fetched.
1058 $key = $db->next_key($key);
1062 Here are some examples of using hashes:
1064 my $db = DBM::Deep->new( "foo.db" );
1066 $db->put("foo", "bar");
1067 print "foo: " . $db->get("foo") . "\n";
1069 $db->put("baz", {}); # new child hash ref
1070 $db->get("baz")->put("buz", "biz");
1071 print "buz: " . $db->get("baz")->get("buz") . "\n";
1073 my $key = $db->first_key();
1075 print "$key: " . $db->get($key) . "\n";
1076 $key = $db->next_key($key);
1079 if ($db->exists("foo")) { $db->delete("foo"); }
1083 For arrays, DBM::Deep supports all the common methods described above, and the
1084 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1085 C<unshift()> and C<splice()>.
1091 Returns the number of elements in the array. Takes no arguments.
1093 my $len = $db->length();
1097 Adds one or more elements onto the end of the array. Accepts scalars, hash
1098 refs or array refs. No return value.
1100 $db->push("foo", "bar", {});
1104 Fetches the last element in the array, and deletes it. Takes no arguments.
1105 Returns undef if array is empty. Returns the element value.
1107 my $elem = $db->pop();
1111 Fetches the first element in the array, deletes it, then shifts all the
1112 remaining elements over to take up the space. Returns the element value. This
1113 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1116 my $elem = $db->shift();
1120 Inserts one or more elements onto the beginning of the array, shifting all
1121 existing elements over to make room. Accepts scalars, hash refs or array refs.
1122 No return value. This method is not recommended with large arrays -- see
1123 <LARGE ARRAYS> below for details.
1125 $db->unshift("foo", "bar", {});
1129 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1130 -f splice> for usage -- it is too complicated to document here. This method is
1131 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1135 Here are some examples of using arrays:
1137 my $db = DBM::Deep->new(
1139 type => DBM::Deep->TYPE_ARRAY
1142 $db->push("bar", "baz");
1143 $db->unshift("foo");
1146 my $len = $db->length();
1147 print "length: $len\n"; # 4
1149 for (my $k=0; $k<$len; $k++) {
1150 print "$k: " . $db->get($k) . "\n";
1153 $db->splice(1, 2, "biz", "baf");
1155 while (my $elem = shift @$db) {
1156 print "shifted: $elem\n";
1161 Enable automatic file locking by passing a true value to the C<locking>
1162 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1164 my $db = DBM::Deep->new(
1169 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1170 mode for writes, and shared mode for reads. This is required if you have
1171 multiple processes accessing the same database file, to avoid file corruption.
1172 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1173 NFS> below for more.
1175 =head2 EXPLICIT LOCKING
1177 You can explicitly lock a database, so it remains locked for multiple
1178 transactions. This is done by calling the C<lock()> method, and passing an
1179 optional lock mode argument (defaults to exclusive mode). This is particularly
1180 useful for things like counters, where the current value needs to be fetched,
1181 then incremented, then stored again.
1184 my $counter = $db->get("counter");
1186 $db->put("counter", $counter);
1195 You can pass C<lock()> an optional argument, which specifies which mode to use
1196 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1197 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1198 same as the constants defined in Perl's C<Fcntl> module.
1200 $db->lock( DBM::Deep->LOCK_SH );
1204 =head1 IMPORTING/EXPORTING
1206 You can import existing complex structures by calling the C<import()> method,
1207 and export an entire database into an in-memory structure using the C<export()>
1208 method. Both are examined here.
1212 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1213 walking the structure and adding keys/elements to the database as you go,
1214 simply pass a reference to the C<import()> method. This recursively adds
1215 everything to an existing DBM::Deep object for you. Here is an example:
1220 array1 => [ "elem0", "elem1", "elem2" ],
1222 subkey1 => "subvalue1",
1223 subkey2 => "subvalue2"
1227 my $db = DBM::Deep->new( "foo.db" );
1228 $db->import( $struct );
1230 print $db->{key1} . "\n"; # prints "value1"
1232 This recursively imports the entire C<$struct> object into C<$db>, including
1233 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1234 keys are merged with the existing ones, replacing if they already exist.
1235 The C<import()> method can be called on any database level (not just the base
1236 level), and works with both hash and array DB types.
1238 B<Note:> Make sure your existing structure has no circular references in it.
1239 These will cause an infinite loop when importing.
1243 Calling the C<export()> method on an existing DBM::Deep object will return
1244 a reference to a new in-memory copy of the database. The export is done
1245 recursively, so all nested hashes/arrays are all exported to standard Perl
1246 objects. Here is an example:
1248 my $db = DBM::Deep->new( "foo.db" );
1250 $db->{key1} = "value1";
1251 $db->{key2} = "value2";
1253 $db->{hash1}->{subkey1} = "subvalue1";
1254 $db->{hash1}->{subkey2} = "subvalue2";
1256 my $struct = $db->export();
1258 print $struct->{key1} . "\n"; # prints "value1"
1260 This makes a complete copy of the database in memory, and returns a reference
1261 to it. The C<export()> method can be called on any database level (not just
1262 the base level), and works with both hash and array DB types. Be careful of
1263 large databases -- you can store a lot more data in a DBM::Deep object than an
1264 in-memory Perl structure.
1266 B<Note:> Make sure your database has no circular references in it.
1267 These will cause an infinite loop when exporting.
1271 DBM::Deep has a number of hooks where you can specify your own Perl function
1272 to perform filtering on incoming or outgoing data. This is a perfect
1273 way to extend the engine, and implement things like real-time compression or
1274 encryption. Filtering applies to the base DB level, and all child hashes /
1275 arrays. Filter hooks can be specified when your DBM::Deep object is first
1276 constructed, or by calling the C<set_filter()> method at any time. There are
1277 four available filter hooks, described below:
1281 =item * filter_store_key
1283 This filter is called whenever a hash key is stored. It
1284 is passed the incoming key, and expected to return a transformed key.
1286 =item * filter_store_value
1288 This filter is called whenever a hash key or array element is stored. It
1289 is passed the incoming value, and expected to return a transformed value.
1291 =item * filter_fetch_key
1293 This filter is called whenever a hash key is fetched (i.e. via
1294 C<first_key()> or C<next_key()>). It is passed the transformed key,
1295 and expected to return the plain key.
1297 =item * filter_fetch_value
1299 This filter is called whenever a hash key or array element is fetched.
1300 It is passed the transformed value, and expected to return the plain value.
1304 Here are the two ways to setup a filter hook:
1306 my $db = DBM::Deep->new(
1308 filter_store_value => \&my_filter_store,
1309 filter_fetch_value => \&my_filter_fetch
1314 $db->set_filter( "filter_store_value", \&my_filter_store );
1315 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1317 Your filter function will be called only when dealing with SCALAR keys or
1318 values. When nested hashes and arrays are being stored/fetched, filtering
1319 is bypassed. Filters are called as static functions, passed a single SCALAR
1320 argument, and expected to return a single SCALAR value. If you want to
1321 remove a filter, set the function reference to C<undef>:
1323 $db->set_filter( "filter_store_value", undef );
1325 =head2 REAL-TIME ENCRYPTION EXAMPLE
1327 Here is a working example that uses the I<Crypt::Blowfish> module to
1328 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1329 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1330 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1333 use Crypt::Blowfish;
1336 my $cipher = Crypt::CBC->new({
1337 'key' => 'my secret key',
1338 'cipher' => 'Blowfish',
1340 'regenerate_key' => 0,
1341 'padding' => 'space',
1345 my $db = DBM::Deep->new(
1346 file => "foo-encrypt.db",
1347 filter_store_key => \&my_encrypt,
1348 filter_store_value => \&my_encrypt,
1349 filter_fetch_key => \&my_decrypt,
1350 filter_fetch_value => \&my_decrypt,
1353 $db->{key1} = "value1";
1354 $db->{key2} = "value2";
1355 print "key1: " . $db->{key1} . "\n";
1356 print "key2: " . $db->{key2} . "\n";
1362 return $cipher->encrypt( $_[0] );
1365 return $cipher->decrypt( $_[0] );
1368 =head2 REAL-TIME COMPRESSION EXAMPLE
1370 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1371 compression / decompression of keys & values with DBM::Deep Filters.
1372 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1373 more on I<Compress::Zlib>.
1378 my $db = DBM::Deep->new(
1379 file => "foo-compress.db",
1380 filter_store_key => \&my_compress,
1381 filter_store_value => \&my_compress,
1382 filter_fetch_key => \&my_decompress,
1383 filter_fetch_value => \&my_decompress,
1386 $db->{key1} = "value1";
1387 $db->{key2} = "value2";
1388 print "key1: " . $db->{key1} . "\n";
1389 print "key2: " . $db->{key2} . "\n";
1395 return Compress::Zlib::memGzip( $_[0] ) ;
1398 return Compress::Zlib::memGunzip( $_[0] ) ;
1401 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1402 actually numerical index numbers, and are not filtered.
1404 =head1 ERROR HANDLING
1406 Most DBM::Deep methods return a true value for success, and call die() on
1407 failure. You can wrap calls in an eval block to catch the die.
1409 my $db = DBM::Deep->new( "foo.db" ); # create hash
1410 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1412 print $@; # prints error message
1414 =head1 LARGEFILE SUPPORT
1416 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1417 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1418 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1419 by calling the static C<set_pack()> method before you do anything else.
1421 DBM::Deep::set_pack(8, 'Q');
1423 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1424 instead of 32-bit longs. After setting these values your DB files have a
1425 theoretical maximum size of 16 XB (exabytes).
1427 B<Note:> Changing these values will B<NOT> work for existing database files.
1428 Only change this for new files, and make sure it stays set consistently
1429 throughout the file's life. If you do set these values, you can no longer
1430 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
1431 back to 32-bit mode.
1433 B<Note:> I have not personally tested files > 2 GB -- all my systems have
1434 only a 32-bit Perl. However, I have received user reports that this does
1437 =head1 LOW-LEVEL ACCESS
1439 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1440 you can call the C<_fh()> method, which returns the handle:
1442 my $fh = $db->_fh();
1444 This method can be called on the root level of the datbase, or any child
1445 hashes or arrays. All levels share a I<root> structure, which contains things
1446 like the filehandle, a reference counter, and all the options specified
1447 when you created the object. You can get access to this root structure by
1448 calling the C<root()> method.
1450 my $root = $db->_root();
1452 This is useful for changing options after the object has already been created,
1453 such as enabling/disabling locking. You can also store your own temporary user
1454 data in this structure (be wary of name collision), which is then accessible from
1455 any child hash or array.
1457 =head1 CUSTOM DIGEST ALGORITHM
1459 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1460 keys. However you can override this, and use another algorithm (such as SHA-256)
1461 or even write your own. But please note that DBM::Deep currently expects zero
1462 collisions, so your algorithm has to be I<perfect>, so to speak.
1463 Collision detection may be introduced in a later version.
1467 You can specify a custom digest algorithm by calling the static C<set_digest()>
1468 function, passing a reference to a subroutine, and the length of the algorithm's
1469 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
1470 objects. Here is a working example that uses a 256-bit hash from the
1471 I<Digest::SHA256> module. Please see
1472 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
1477 my $context = Digest::SHA256::new(256);
1479 DBM::Deep::set_digest( \&my_digest, 32 );
1481 my $db = DBM::Deep->new( "foo-sha.db" );
1483 $db->{key1} = "value1";
1484 $db->{key2} = "value2";
1485 print "key1: " . $db->{key1} . "\n";
1486 print "key2: " . $db->{key2} . "\n";
1492 return substr( $context->hash($_[0]), 0, 32 );
1495 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1496 of bytes you specify in the C<set_digest()> function (in this case 32).
1498 =head1 CIRCULAR REFERENCES
1500 DBM::Deep has B<experimental> support for circular references. Meaning you
1501 can have a nested hash key or array element that points to a parent object.
1502 This relationship is stored in the DB file, and is preserved between sessions.
1505 my $db = DBM::Deep->new( "foo.db" );
1508 $db->{circle} = $db; # ref to self
1510 print $db->{foo} . "\n"; # prints "foo"
1511 print $db->{circle}->{foo} . "\n"; # prints "foo" again
1513 B<Note>: Passing the object to a function that recursively walks the
1514 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1515 C<export()> methods) will result in an infinite loop. This will be fixed in
1518 =head1 CAVEATS / ISSUES / BUGS
1520 This section describes all the known issues with DBM::Deep. It you have found
1521 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1523 =head2 UNUSED SPACE RECOVERY
1525 One major caveat with DBM::Deep is that space occupied by existing keys and
1526 values is not recovered when they are deleted. Meaning if you keep deleting
1527 and adding new keys, your file will continuously grow. I am working on this,
1528 but in the meantime you can call the built-in C<optimize()> method from time to
1529 time (perhaps in a crontab or something) to recover all your unused space.
1531 $db->optimize(); # returns true on success
1533 This rebuilds the ENTIRE database into a new file, then moves it on top of
1534 the original. The new file will have no unused space, thus it will take up as
1535 little disk space as possible. Please note that this operation can take
1536 a long time for large files, and you need enough disk space to temporarily hold
1537 2 copies of your DB file. The temporary file is created in the same directory
1538 as the original, named with a ".tmp" extension, and is deleted when the
1539 operation completes. Oh, and if locking is enabled, the DB is automatically
1540 locked for the entire duration of the copy.
1542 B<WARNING:> Only call optimize() on the top-level node of the database, and
1543 make sure there are no child references lying around. DBM::Deep keeps a reference
1544 counter, and if it is greater than 1, optimize() will abort and return undef.
1546 =head2 AUTOVIVIFICATION
1548 Unfortunately, autovivification doesn't work with tied hashes. This appears to
1549 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
1550 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
1551 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
1552 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
1555 $db->{foo}->{bar} = "hello";
1557 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
1558 being an empty hash. Try this instead, which works fine:
1560 $db->{foo} = { bar => "hello" };
1562 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
1563 the execution path, and Perl indeed passes an empty hash to the STORE() method.
1564 Probably a bug in Perl.
1568 (The reasons given assume a high level of Perl understanding, specifically of
1569 references. You can safely skip this section.)
1571 Currently, the only references supported are HASH and ARRAY. The other reference
1572 types (SCALAR, CODE, GLOB, and REF) cannot be supported for various reasons.
1578 These are things like filehandles and other sockets. They can't be supported
1579 because it's completely unclear how DBM::Deep should serialize them.
1581 =item * SCALAR / REF
1583 The discussion here refers to the following type of example:
1590 # In some other process ...
1592 my $val = ${ $db->{key1} };
1594 is( $val, 50, "What actually gets stored in the DB file?" );
1596 The problem is one of synchronization. When the variable being referred to
1597 changes value, the reference isn't notified. This means that the new value won't
1598 be stored in the datafile for other processes to read. There is no TIEREF.
1600 It is theoretically possible to store references to values already within a
1601 DBM::Deep object because everything already is synchronized, but the change to
1602 the internals would be quite large. Specifically, DBM::Deep would have to tie
1603 every single value that is stored. This would bloat the RAM footprint of
1604 DBM::Deep at least twofold (if not more) and be a significant performance drain,
1605 all to support a feature that has never been requested.
1609 L<http://search.cpan.org/search?module=Data::Dump::Streamer> provides a
1610 mechanism for serializing coderefs, including saving off all closure state.
1611 However, just as for SCALAR and REF, that closure state may change without
1612 notifying the DBM::Deep object storing the reference.
1616 =head2 FILE CORRUPTION
1618 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1619 for a 32-bit signature when opened, but other corruption in files can cause
1620 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1621 stuck in an infinite loop depending on the level of corruption. File write
1622 operations are not checked for failure (for speed), so if you happen to run
1623 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1624 be addressed in a later version of DBM::Deep.
1628 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1629 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1630 about setting up your NFS server with a locking daemon, then using lockf() to
1631 lock your files, but your mileage may vary there as well. From what I
1632 understand, there is no real way to do it. However, if you need access to the
1633 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1634 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1636 =head2 COPYING OBJECTS
1638 Beware of copying tied objects in Perl. Very strange things can happen.
1639 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1640 returns a new, blessed, tied hash or array to the same level in the DB.
1642 my $copy = $db->clone();
1644 B<Note>: Since clone() here is cloning the object, not the database location, any
1645 modifications to either $db or $copy will be visible in both.
1649 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1650 These functions cause every element in the array to move, which can be murder
1651 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1652 a different location. This will be addressed in the forthcoming version 1.00.
1654 =head2 WRITEONLY FILES
1656 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1657 writeonly mode. STORE will verify that the filehandle is writable. However, there
1658 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1659 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1663 This section discusses DBM::Deep's speed and memory usage.
1667 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1668 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1669 multi-level hash/array support, and cross-platform FTPable files. Even so,
1670 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1671 with huge databases. Here is some test data:
1673 Adding 1,000,000 keys to new DB file...
1675 At 100 keys, avg. speed is 2,703 keys/sec
1676 At 200 keys, avg. speed is 2,642 keys/sec
1677 At 300 keys, avg. speed is 2,598 keys/sec
1678 At 400 keys, avg. speed is 2,578 keys/sec
1679 At 500 keys, avg. speed is 2,722 keys/sec
1680 At 600 keys, avg. speed is 2,628 keys/sec
1681 At 700 keys, avg. speed is 2,700 keys/sec
1682 At 800 keys, avg. speed is 2,607 keys/sec
1683 At 900 keys, avg. speed is 2,190 keys/sec
1684 At 1,000 keys, avg. speed is 2,570 keys/sec
1685 At 2,000 keys, avg. speed is 2,417 keys/sec
1686 At 3,000 keys, avg. speed is 1,982 keys/sec
1687 At 4,000 keys, avg. speed is 1,568 keys/sec
1688 At 5,000 keys, avg. speed is 1,533 keys/sec
1689 At 6,000 keys, avg. speed is 1,787 keys/sec
1690 At 7,000 keys, avg. speed is 1,977 keys/sec
1691 At 8,000 keys, avg. speed is 2,028 keys/sec
1692 At 9,000 keys, avg. speed is 2,077 keys/sec
1693 At 10,000 keys, avg. speed is 2,031 keys/sec
1694 At 20,000 keys, avg. speed is 1,970 keys/sec
1695 At 30,000 keys, avg. speed is 2,050 keys/sec
1696 At 40,000 keys, avg. speed is 2,073 keys/sec
1697 At 50,000 keys, avg. speed is 1,973 keys/sec
1698 At 60,000 keys, avg. speed is 1,914 keys/sec
1699 At 70,000 keys, avg. speed is 2,091 keys/sec
1700 At 80,000 keys, avg. speed is 2,103 keys/sec
1701 At 90,000 keys, avg. speed is 1,886 keys/sec
1702 At 100,000 keys, avg. speed is 1,970 keys/sec
1703 At 200,000 keys, avg. speed is 2,053 keys/sec
1704 At 300,000 keys, avg. speed is 1,697 keys/sec
1705 At 400,000 keys, avg. speed is 1,838 keys/sec
1706 At 500,000 keys, avg. speed is 1,941 keys/sec
1707 At 600,000 keys, avg. speed is 1,930 keys/sec
1708 At 700,000 keys, avg. speed is 1,735 keys/sec
1709 At 800,000 keys, avg. speed is 1,795 keys/sec
1710 At 900,000 keys, avg. speed is 1,221 keys/sec
1711 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1713 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1714 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1715 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1716 Run time was 12 min 3 sec.
1720 One of the great things about DBM::Deep is that it uses very little memory.
1721 Even with huge databases (1,000,000+ keys) you will not see much increased
1722 memory on your process. DBM::Deep relies solely on the filesystem for storing
1723 and fetching data. Here is output from I</usr/bin/top> before even opening a
1726 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1727 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1729 Basically the process is taking 2,716K of memory. And here is the same
1730 process after storing and fetching 1,000,000 keys:
1732 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1733 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1735 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1736 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1738 =head1 DB FILE FORMAT
1740 In case you were interested in the underlying DB file format, it is documented
1741 here in this section. You don't need to know this to use the module, it's just
1742 included for reference.
1746 DBM::Deep files always start with a 32-bit signature to identify the file type.
1747 This is at offset 0. The signature is "DPDB" in network byte order. This is
1748 checked for when the file is opened and an error will be thrown if it's not found.
1752 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
1753 has a standard header containing the type of data, the length of data, and then
1754 the data itself. The type is a single character (1 byte), the length is a
1755 32-bit unsigned long in network byte order, and the data is, well, the data.
1756 Here is how it unfolds:
1760 Immediately after the 32-bit file signature is the I<Master Index> record.
1761 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
1762 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
1763 depending on how the DBM::Deep object was constructed.
1765 The index works by looking at a I<MD5 Hash> of the hash key (or array index
1766 number). The first 8-bit char of the MD5 signature is the offset into the
1767 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
1768 index element is a file offset of the next tag for the key/element in question,
1769 which is usually a I<Bucket List> tag (see below).
1771 The next tag I<could> be another index, depending on how many keys/elements
1772 exist. See L<RE-INDEXING> below for details.
1776 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
1777 file offsets to where the actual data is stored. It starts with a standard
1778 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
1779 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
1780 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
1781 When the list fills up, a I<Re-Index> operation is performed (See
1782 L<RE-INDEXING> below).
1786 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
1787 index/value pair (in array mode). It starts with a standard tag header with
1788 type I<D> for scalar data (string, binary, etc.), or it could be a nested
1789 hash (type I<H>) or array (type I<A>). The value comes just after the tag
1790 header. The size reported in the tag header is only for the value, but then,
1791 just after the value is another size (32-bit unsigned long) and then the plain
1792 key itself. Since the value is likely to be fetched more often than the plain
1793 key, I figured it would be I<slightly> faster to store the value first.
1795 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
1796 record for the nested structure, where the process begins all over again.
1800 After a I<Bucket List> grows to 16 records, its allocated space in the file is
1801 exhausted. Then, when another key/element comes in, the list is converted to a
1802 new index record. However, this index will look at the next char in the MD5
1803 hash, and arrange new Bucket List pointers accordingly. This process is called
1804 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
1805 17 (16 + new one) keys/elements are removed from the old Bucket List and
1806 inserted into the new index. Several new Bucket Lists are created in the
1807 process, as a new MD5 char from the key is being examined (it is unlikely that
1808 the keys will all share the same next char of their MD5s).
1810 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
1811 when the Bucket Lists will turn into indexes, but the first round tends to
1812 happen right around 4,000 keys. You will see a I<slight> decrease in
1813 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
1814 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
1815 right around 900,000 keys. This process can continue nearly indefinitely --
1816 right up until the point the I<MD5> signatures start colliding with each other,
1817 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
1818 getting struck by lightning while you are walking to cash in your tickets.
1819 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
1820 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
1821 this is 340 unodecillion, but don't quote me).
1825 When a new key/element is stored, the key (or index number) is first run through
1826 I<Digest::MD5> to get a 128-bit signature (example, in hex:
1827 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
1828 for the first char of the signature (in this case I<b0>). If it does not exist,
1829 a new I<Bucket List> is created for our key (and the next 15 future keys that
1830 happen to also have I<b> as their first MD5 char). The entire MD5 is written
1831 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
1832 this point, unless we are replacing an existing I<Bucket>), where the actual
1833 data will be stored.
1837 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
1838 (or index number), then walking along the indexes. If there are enough
1839 keys/elements in this DB level, there might be nested indexes, each linked to
1840 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
1841 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
1842 question. If we found a match, the I<Bucket> tag is loaded, where the value and
1843 plain key are stored.
1845 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
1846 methods. In this process the indexes are walked systematically, and each key
1847 fetched in increasing MD5 order (which is why it appears random). Once the
1848 I<Bucket> is found, the value is skipped and the plain key returned instead.
1849 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
1850 alphabetically sorted. This only happens on an index-level -- as soon as the
1851 I<Bucket Lists> are hit, the keys will come out in the order they went in --
1852 so it's pretty much undefined how the keys will come out -- just like Perl's
1855 =head1 CODE COVERAGE
1857 We use B<Devel::Cover> to test the code coverage of our tests, below is the
1858 B<Devel::Cover> report on this module's test suite.
1860 ----------------------------------- ------ ------ ------ ------ ------ ------
1861 File stmt bran cond sub time total
1862 ----------------------------------- ------ ------ ------ ------ ------ ------
1863 blib/lib/DBM/Deep.pm 94.9 80.6 73.0 100.0 37.9 90.4
1864 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 18.2 98.1
1865 blib/lib/DBM/Deep/Engine.pm 98.9 87.3 80.0 100.0 34.2 95.2
1866 blib/lib/DBM/Deep/Hash.pm 100.0 87.5 100.0 100.0 9.7 97.3
1867 Total 97.9 85.9 79.7 100.0 100.0 94.3
1868 ----------------------------------- ------ ------ ------ ------ ------ ------
1870 =head1 MORE INFORMATION
1872 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
1873 or send email to L<DBM-Deep@googlegroups.com>.
1877 Joseph Huckaby, L<jhuckaby@cpan.org>
1879 Rob Kinyon, L<rkinyon@cpan.org>
1881 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
1885 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
1886 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
1890 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
1891 This is free software, you may use it and distribute it under the
1892 same terms as Perl itself.