7 # Multi-level database module for storing hash trees, arrays and simple
8 # key/value pairs into FTP-able, cross-platform binary database files.
10 # Type `perldoc DBM::Deep` for complete documentation.
14 # tie %db, 'DBM::Deep', 'my_database.db'; # standard tie() method
16 # my $db = new DBM::Deep( 'my_database.db' ); # preferred OO method
18 # $db->{my_scalar} = 'hello world';
19 # $db->{my_hash} = { larry => 'genius', hashes => 'fast' };
20 # $db->{my_array} = [ 1, 2, 3, time() ];
21 # $db->{my_complex} = [ 'hello', { perl => 'rules' }, 42, 99 ];
22 # push @{$db->{my_array}}, 'another value';
23 # my @key_list = keys %{$db->{my_hash}};
24 # print "This module " . $db->{my_complex}->[1]->{perl} . "!\n";
27 # (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
28 # This program is free software; you can redistribute it and/or
29 # modify it under the same terms as Perl itself.
34 use Fcntl qw( :DEFAULT :flock :seek );
38 use DBM::Deep::Engine;
40 use vars qw( $VERSION );
41 $VERSION = q(0.99_01);
44 # Setup constants for users to pass to new()
46 sub TYPE_HASH () { DBM::Deep::Engine->SIG_HASH }
47 sub TYPE_ARRAY () { DBM::Deep::Engine->SIG_ARRAY }
55 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
60 unless ( eval { local $SIG{'__DIE__'}; %{$_[0]} || 1 } ) {
61 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
66 $args = { file => shift };
74 # Class constructor method for Perl OO interface.
75 # Calls tie() and returns blessed reference to tied hash or array,
76 # providing a hybrid OO/tie interface.
79 my $args = $class->_get_args( @_ );
82 # Check if we want a tied hash or array.
85 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
86 $class = 'DBM::Deep::Array';
87 require DBM::Deep::Array;
88 tie @$self, $class, %$args;
91 $class = 'DBM::Deep::Hash';
92 require DBM::Deep::Hash;
93 tie %$self, $class, %$args;
96 return bless $self, $class;
101 # Setup $self and bless into this class.
106 # These are the defaults to be optionally overridden below
109 engine => DBM::Deep::Engine->new,
110 base_offset => undef,
113 foreach my $param ( keys %$self ) {
114 next unless exists $args->{$param};
115 $self->{$param} = delete $args->{$param}
118 # locking implicitly enables autoflush
119 if ($args->{locking}) { $args->{autoflush} = 1; }
121 $self->{root} = exists $args->{root}
123 : DBM::Deep::_::Root->new( $args );
125 $self->{engine}->setup_fh( $self );
132 require DBM::Deep::Hash;
133 return DBM::Deep::Hash->TIEHASH( @_ );
138 require DBM::Deep::Array;
139 return DBM::Deep::Array->TIEARRAY( @_ );
142 #XXX Unneeded now ...
148 # If db locking is set, flock() the db file. If called multiple
149 # times before unlock(), then the same number of unlocks() must
150 # be called before the lock is released.
152 my $self = shift->_get_self;
154 $type = LOCK_EX unless defined $type;
156 if (!defined($self->_fh)) { return; }
158 if ($self->_root->{locking}) {
159 if (!$self->_root->{locked}) {
160 flock($self->_fh, $type);
162 # refresh end counter in case file has changed size
163 my @stats = stat($self->_fh);
164 $self->_root->{end} = $stats[7];
166 # double-check file inode, in case another process
167 # has optimize()d our file while we were waiting.
168 if ($stats[1] != $self->_root->{inode}) {
169 $self->{engine}->close_fh( $self );
170 $self->{engine}->setup_fh( $self );
171 flock($self->_fh, $type); # re-lock
173 # This may not be necessary after re-opening
174 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
177 $self->_root->{locked}++;
187 # If db locking is set, unlock the db file. See note in lock()
188 # regarding calling lock() multiple times.
190 my $self = shift->_get_self;
192 if (!defined($self->_fh)) { return; }
194 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
195 $self->_root->{locked}--;
196 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
205 my $self = shift->_get_self;
206 my ($spot, $value) = @_;
211 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
212 ${$spot} = $value->_repr;
213 $value->_copy_node( ${$spot} );
216 my $r = Scalar::Util::reftype( $value );
217 my $c = Scalar::Util::blessed( $value );
218 if ( $r eq 'ARRAY' ) {
219 ${$spot} = [ @{$value} ];
222 ${$spot} = { %{$value} };
224 ${$spot} = bless ${$spot}, $c
232 die "Must be implemented in a child class\n";
236 die "Must be implemented in a child class\n";
241 # Recursively export into standard Perl hashes and arrays.
243 my $self = shift->_get_self;
245 my $temp = $self->_repr;
248 $self->_copy_node( $temp );
256 # Recursively import Perl hash/array structure
258 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
260 my $self = shift->_get_self;
263 # struct is not a reference, so just import based on our type
265 $struct = $self->_repr( @_ );
268 return $self->_import( $struct );
273 # Rebuild entire database into new file, then move
274 # it back on top of original.
276 my $self = shift->_get_self;
278 #XXX Need to create a new test for this
279 # if ($self->_root->{links} > 1) {
280 # $self->_throw_error("Cannot optimize: reference count is greater than 1");
283 my $db_temp = DBM::Deep->new(
284 file => $self->_root->{file} . '.tmp',
289 $self->_copy_node( $db_temp );
293 # Attempt to copy user, group and permissions over to new file
295 my @stats = stat($self->_fh);
296 my $perms = $stats[2] & 07777;
299 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
300 chmod( $perms, $self->_root->{file} . '.tmp' );
302 # q.v. perlport for more information on this variable
303 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
305 # Potential race condition when optmizing on Win32 with locking.
306 # The Windows filesystem requires that the filehandle be closed
307 # before it is overwritten with rename(). This could be redone
311 $self->{engine}->close_fh( $self );
314 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
315 unlink $self->_root->{file} . '.tmp';
317 $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
321 $self->{engine}->close_fh( $self );
322 $self->{engine}->setup_fh( $self );
329 # Make copy of object and return
331 my $self = shift->_get_self;
333 return DBM::Deep->new(
334 type => $self->_type,
335 base_offset => $self->_base_offset,
341 my %is_legal_filter = map {
344 store_key store_value
345 fetch_key fetch_value
350 # Setup filter function for storing or fetching the key or value
352 my $self = shift->_get_self;
356 if ( $is_legal_filter{$type} ) {
357 $self->_root->{"filter_$type"} = $func;
371 # Get access to the root structure
373 my $self = $_[0]->_get_self;
374 return $self->{root};
379 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
381 my $self = $_[0]->_get_self;
382 return $self->{type};
387 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
389 my $self = $_[0]->_get_self;
390 return $self->{base_offset};
395 # Get access to the raw fh
397 my $self = $_[0]->_get_self;
398 return $self->_root->{fh};
406 die "DBM::Deep: $_[1]\n";
411 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
416 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
421 # Store single hash key/value or array element in database.
423 my $self = shift->_get_self;
424 my ($key, $value) = @_;
426 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
427 $self->_throw_error( 'Cannot write to a readonly filehandle' );
431 # Request exclusive lock for writing
433 $self->lock( LOCK_EX );
435 my $md5 = $self->{engine}{digest}->($key);
437 my $tag = $self->{engine}->find_bucket_list( $self, $md5, { create => 1 } );
439 # User may be storing a hash, in which case we do not want it run
440 # through the filtering system
441 if ( !ref($value) && $self->_root->{filter_store_value} ) {
442 $value = $self->_root->{filter_store_value}->( $value );
446 # Add key/value to bucket list
448 my $result = $self->{engine}->add_bucket( $self, $tag, $md5, $key, $value );
457 # Fetch single value or element given plain key or array index
459 my $self = shift->_get_self;
462 my $md5 = $self->{engine}{digest}->($key);
465 # Request shared lock for reading
467 $self->lock( LOCK_SH );
469 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
476 # Get value from bucket list
478 my $result = $self->{engine}->get_bucket_value( $self, $tag, $md5 );
482 # Filters only apply to scalar values, so the ref check is making
483 # sure the fetched bucket is a scalar, not a child hash or array.
484 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
485 ? $self->_root->{filter_fetch_value}->($result)
491 # Delete single key/value pair or element given plain key or array index
493 my $self = $_[0]->_get_self;
496 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
497 $self->_throw_error( 'Cannot write to a readonly filehandle' );
501 # Request exclusive lock for writing
503 $self->lock( LOCK_EX );
505 my $md5 = $self->{engine}{digest}->($key);
507 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
516 my $value = $self->{engine}->get_bucket_value($self, $tag, $md5 );
518 if (defined $value && !ref($value) && $self->_root->{filter_fetch_value}) {
519 $value = $self->_root->{filter_fetch_value}->($value);
522 my $result = $self->{engine}->delete_bucket( $self, $tag, $md5 );
525 # If this object is an array and the key deleted was on the end of the stack,
526 # decrement the length variable.
536 # Check if a single key or element exists given plain key or array index
538 my $self = $_[0]->_get_self;
541 my $md5 = $self->{engine}{digest}->($key);
544 # Request shared lock for reading
546 $self->lock( LOCK_SH );
548 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
553 # For some reason, the built-in exists() function returns '' for false
559 # Check if bucket exists and return 1 or ''
561 my $result = $self->{engine}->bucket_exists( $self, $tag, $md5 ) || '';
570 # Clear all keys from hash, or all elements from array.
572 my $self = $_[0]->_get_self;
574 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
575 $self->_throw_error( 'Cannot write to a readonly filehandle' );
579 # Request exclusive lock for writing
581 $self->lock( LOCK_EX );
585 seek($fh, $self->_base_offset + $self->_root->{file_offset}, SEEK_SET);
591 #XXX This needs updating to use _release_space
592 $self->{engine}->write_tag(
593 $self, $self->_base_offset, $self->_type,
594 chr(0)x$self->{engine}{index_size},
603 # Public method aliases
605 sub put { (shift)->STORE( @_ ) }
606 sub store { (shift)->STORE( @_ ) }
607 sub get { (shift)->FETCH( @_ ) }
608 sub fetch { (shift)->FETCH( @_ ) }
609 sub delete { (shift)->DELETE( @_ ) }
610 sub exists { (shift)->EXISTS( @_ ) }
611 sub clear { (shift)->CLEAR( @_ ) }
613 package DBM::Deep::_::Root;
628 filter_store_key => undef,
629 filter_store_value => undef,
630 filter_fetch_key => undef,
631 filter_fetch_value => undef,
635 if ( $self->{fh} && !$self->{file_offset} ) {
636 $self->{file_offset} = tell( $self->{fh} );
646 close $self->{fh} if $self->{fh};
656 DBM::Deep - A pure perl multi-level hash/array DBM
661 my $db = DBM::Deep->new( "foo.db" );
663 $db->{key} = 'value'; # tie() style
666 $db->put('key' => 'value'); # OO style
667 print $db->get('key');
669 # true multi-level support
670 $db->{my_complex} = [
671 'hello', { perl => 'rules' },
677 A unique flat-file database module, written in pure perl. True
678 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
679 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
680 handle millions of keys and unlimited hash levels without significant
681 slow-down. Written from the ground-up in pure perl -- this is NOT a
682 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
683 Mac OS X and Windows.
685 =head1 VERSION DIFFERENCES
687 B<NOTE>: 0.99_01 and above have significant file format differences from 0.98 and
688 before. While attempts have been made to be backwards compatible, no guarantees.
692 Hopefully you are using Perl's excellent CPAN module, which will download
693 and install the module for you. If not, get the tarball, and run these
705 Construction can be done OO-style (which is the recommended way), or using
706 Perl's tie() function. Both are examined here.
708 =head2 OO CONSTRUCTION
710 The recommended way to construct a DBM::Deep object is to use the new()
711 method, which gets you a blessed, tied hash or array reference.
713 my $db = DBM::Deep->new( "foo.db" );
715 This opens a new database handle, mapped to the file "foo.db". If this
716 file does not exist, it will automatically be created. DB files are
717 opened in "r+" (read/write) mode, and the type of object returned is a
718 hash, unless otherwise specified (see L<OPTIONS> below).
720 You can pass a number of options to the constructor to specify things like
721 locking, autoflush, etc. This is done by passing an inline hash:
723 my $db = DBM::Deep->new(
729 Notice that the filename is now specified I<inside> the hash with
730 the "file" parameter, as opposed to being the sole argument to the
731 constructor. This is required if any options are specified.
732 See L<OPTIONS> below for the complete list.
736 You can also start with an array instead of a hash. For this, you must
737 specify the C<type> parameter:
739 my $db = DBM::Deep->new(
741 type => DBM::Deep->TYPE_ARRAY
744 B<Note:> Specifing the C<type> parameter only takes effect when beginning
745 a new DB file. If you create a DBM::Deep object with an existing file, the
746 C<type> will be loaded from the file header, and an error will be thrown if
747 the wrong type is passed in.
749 =head2 TIE CONSTRUCTION
751 Alternately, you can create a DBM::Deep handle by using Perl's built-in
752 tie() function. The object returned from tie() can be used to call methods,
753 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
754 file (as expected with most tie'd objects).
757 my $db = tie %hash, "DBM::Deep", "foo.db";
760 my $db = tie @array, "DBM::Deep", "bar.db";
762 As with the OO constructor, you can replace the DB filename parameter with
763 a hash containing one or more options (see L<OPTIONS> just below for the
766 tie %hash, "DBM::Deep", {
774 There are a number of options that can be passed in when constructing your
775 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
781 Filename of the DB file to link the handle to. You can pass a full absolute
782 filesystem path, partial path, or a plain filename if the file is in the
783 current working directory. This is a required parameter (though q.v. fh).
787 If you want, you can pass in the fh instead of the file. This is most useful for doing
790 my $db = DBM::Deep->new( { fh => \*DATA } );
792 You are responsible for making sure that the fh has been opened appropriately for your
793 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
794 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
795 needs to read from the fh.
799 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
800 not need to set this. However, it's there if you want it.
802 If you pass in fh and do not set this, it will be set appropriately.
806 This parameter specifies what type of object to create, a hash or array. Use
807 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
808 This only takes effect when beginning a new file. This is an optional
809 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
813 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
814 function to lock the database in exclusive mode for writes, and shared mode for
815 reads. Pass any true value to enable. This affects the base DB handle I<and
816 any child hashes or arrays> that use the same DB file. This is an optional
817 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
821 Specifies whether autoflush is to be enabled on the underlying filehandle.
822 This obviously slows down write operations, but is required if you may have
823 multiple processes accessing the same DB file (also consider enable I<locking>).
824 Pass any true value to enable. This is an optional parameter, and defaults to 0
829 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
830 restore them when fetched. This is an B<experimental> feature, and does have
831 side-effects. Basically, when hashes are re-blessed into their original
832 classes, they are no longer blessed into the DBM::Deep class! So you won't be
833 able to call any DBM::Deep methods on them. You have been warned.
834 This is an optional parameter, and defaults to 0 (disabled).
838 See L<FILTERS> below.
844 With DBM::Deep you can access your databases using Perl's standard hash/array
845 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
846 treat them as such. DBM::Deep will intercept all reads/writes and direct them
847 to the right place -- the DB file. This has nothing to do with the
848 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
849 using regular hashes and arrays, rather than calling functions like C<get()>
850 and C<put()> (although those work too). It is entirely up to you how to want
851 to access your databases.
855 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
856 or even nested hashes (or arrays) using standard Perl syntax:
858 my $db = DBM::Deep->new( "foo.db" );
860 $db->{mykey} = "myvalue";
862 $db->{myhash}->{subkey} = "subvalue";
864 print $db->{myhash}->{subkey} . "\n";
866 You can even step through hash keys using the normal Perl C<keys()> function:
868 foreach my $key (keys %$db) {
869 print "$key: " . $db->{$key} . "\n";
872 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
873 pushes them onto an array, all before the loop even begins. If you have an
874 extra large hash, this may exhaust Perl's memory. Instead, consider using
875 Perl's C<each()> function, which pulls keys/values one at a time, using very
878 while (my ($key, $value) = each %$db) {
879 print "$key: $value\n";
882 Please note that when using C<each()>, you should always pass a direct
883 hash reference, not a lookup. Meaning, you should B<never> do this:
886 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
888 This causes an infinite loop, because for each iteration, Perl is calling
889 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
890 it effectively keeps returning the first key over and over again. Instead,
891 assign a temporary variable to C<$db->{foo}>, then pass that to each().
895 As with hashes, you can treat any DBM::Deep object like a normal Perl array
896 reference. This includes inserting, removing and manipulating elements,
897 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
898 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
899 or simply be a nested array reference inside a hash. Example:
901 my $db = DBM::Deep->new(
902 file => "foo-array.db",
903 type => DBM::Deep->TYPE_ARRAY
907 push @$db, "bar", "baz";
910 my $last_elem = pop @$db; # baz
911 my $first_elem = shift @$db; # bah
912 my $second_elem = $db->[1]; # bar
914 my $num_elements = scalar @$db;
918 In addition to the I<tie()> interface, you can also use a standard OO interface
919 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
920 array) has its own methods, but both types share the following common methods:
921 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
925 =item * new() / clone()
927 These are the constructor and copy-functions.
929 =item * put() / store()
931 Stores a new hash key/value pair, or sets an array element value. Takes two
932 arguments, the hash key or array index, and the new value. The value can be
933 a scalar, hash ref or array ref. Returns true on success, false on failure.
935 $db->put("foo", "bar"); # for hashes
936 $db->put(1, "bar"); # for arrays
938 =item * get() / fetch()
940 Fetches the value of a hash key or array element. Takes one argument: the hash
941 key or array index. Returns a scalar, hash ref or array ref, depending on the
944 my $value = $db->get("foo"); # for hashes
945 my $value = $db->get(1); # for arrays
949 Checks if a hash key or array index exists. Takes one argument: the hash key
950 or array index. Returns true if it exists, false if not.
952 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
953 if ($db->exists(1)) { print "yay!\n"; } # for arrays
957 Deletes one hash key/value pair or array element. Takes one argument: the hash
958 key or array index. Returns true on success, false if not found. For arrays,
959 the remaining elements located after the deleted element are NOT moved over.
960 The deleted element is essentially just undefined, which is exactly how Perl's
961 internal arrays work. Please note that the space occupied by the deleted
962 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
963 below for details and workarounds.
965 $db->delete("foo"); # for hashes
966 $db->delete(1); # for arrays
970 Deletes B<all> hash keys or array elements. Takes no arguments. No return
971 value. Please note that the space occupied by the deleted keys/values or
972 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
973 details and workarounds.
975 $db->clear(); # hashes or arrays
977 =item * lock() / unlock()
983 Recover lost disk space.
985 =item * import() / export()
987 Data going in and out.
989 =item * set_digest() / set_pack() / set_filter()
991 q.v. adjusting the interal parameters.
997 For hashes, DBM::Deep supports all the common methods described above, and the
998 following additional methods: C<first_key()> and C<next_key()>.
1004 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1005 fetched in an undefined order (which appears random). Takes no arguments,
1006 returns the key as a scalar value.
1008 my $key = $db->first_key();
1012 Returns the "next" key in the hash, given the previous one as the sole argument.
1013 Returns undef if there are no more keys to be fetched.
1015 $key = $db->next_key($key);
1019 Here are some examples of using hashes:
1021 my $db = DBM::Deep->new( "foo.db" );
1023 $db->put("foo", "bar");
1024 print "foo: " . $db->get("foo") . "\n";
1026 $db->put("baz", {}); # new child hash ref
1027 $db->get("baz")->put("buz", "biz");
1028 print "buz: " . $db->get("baz")->get("buz") . "\n";
1030 my $key = $db->first_key();
1032 print "$key: " . $db->get($key) . "\n";
1033 $key = $db->next_key($key);
1036 if ($db->exists("foo")) { $db->delete("foo"); }
1040 For arrays, DBM::Deep supports all the common methods described above, and the
1041 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1042 C<unshift()> and C<splice()>.
1048 Returns the number of elements in the array. Takes no arguments.
1050 my $len = $db->length();
1054 Adds one or more elements onto the end of the array. Accepts scalars, hash
1055 refs or array refs. No return value.
1057 $db->push("foo", "bar", {});
1061 Fetches the last element in the array, and deletes it. Takes no arguments.
1062 Returns undef if array is empty. Returns the element value.
1064 my $elem = $db->pop();
1068 Fetches the first element in the array, deletes it, then shifts all the
1069 remaining elements over to take up the space. Returns the element value. This
1070 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1073 my $elem = $db->shift();
1077 Inserts one or more elements onto the beginning of the array, shifting all
1078 existing elements over to make room. Accepts scalars, hash refs or array refs.
1079 No return value. This method is not recommended with large arrays -- see
1080 <LARGE ARRAYS> below for details.
1082 $db->unshift("foo", "bar", {});
1086 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1087 -f splice> for usage -- it is too complicated to document here. This method is
1088 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1092 Here are some examples of using arrays:
1094 my $db = DBM::Deep->new(
1096 type => DBM::Deep->TYPE_ARRAY
1099 $db->push("bar", "baz");
1100 $db->unshift("foo");
1103 my $len = $db->length();
1104 print "length: $len\n"; # 4
1106 for (my $k=0; $k<$len; $k++) {
1107 print "$k: " . $db->get($k) . "\n";
1110 $db->splice(1, 2, "biz", "baf");
1112 while (my $elem = shift @$db) {
1113 print "shifted: $elem\n";
1118 Enable automatic file locking by passing a true value to the C<locking>
1119 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1121 my $db = DBM::Deep->new(
1126 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1127 mode for writes, and shared mode for reads. This is required if you have
1128 multiple processes accessing the same database file, to avoid file corruption.
1129 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1130 NFS> below for more.
1132 =head2 EXPLICIT LOCKING
1134 You can explicitly lock a database, so it remains locked for multiple
1135 transactions. This is done by calling the C<lock()> method, and passing an
1136 optional lock mode argument (defaults to exclusive mode). This is particularly
1137 useful for things like counters, where the current value needs to be fetched,
1138 then incremented, then stored again.
1141 my $counter = $db->get("counter");
1143 $db->put("counter", $counter);
1152 You can pass C<lock()> an optional argument, which specifies which mode to use
1153 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1154 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1155 same as the constants defined in Perl's C<Fcntl> module.
1157 $db->lock( DBM::Deep->LOCK_SH );
1161 =head1 IMPORTING/EXPORTING
1163 You can import existing complex structures by calling the C<import()> method,
1164 and export an entire database into an in-memory structure using the C<export()>
1165 method. Both are examined here.
1169 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1170 walking the structure and adding keys/elements to the database as you go,
1171 simply pass a reference to the C<import()> method. This recursively adds
1172 everything to an existing DBM::Deep object for you. Here is an example:
1177 array1 => [ "elem0", "elem1", "elem2" ],
1179 subkey1 => "subvalue1",
1180 subkey2 => "subvalue2"
1184 my $db = DBM::Deep->new( "foo.db" );
1185 $db->import( $struct );
1187 print $db->{key1} . "\n"; # prints "value1"
1189 This recursively imports the entire C<$struct> object into C<$db>, including
1190 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1191 keys are merged with the existing ones, replacing if they already exist.
1192 The C<import()> method can be called on any database level (not just the base
1193 level), and works with both hash and array DB types.
1195 B<Note:> Make sure your existing structure has no circular references in it.
1196 These will cause an infinite loop when importing.
1200 Calling the C<export()> method on an existing DBM::Deep object will return
1201 a reference to a new in-memory copy of the database. The export is done
1202 recursively, so all nested hashes/arrays are all exported to standard Perl
1203 objects. Here is an example:
1205 my $db = DBM::Deep->new( "foo.db" );
1207 $db->{key1} = "value1";
1208 $db->{key2} = "value2";
1210 $db->{hash1}->{subkey1} = "subvalue1";
1211 $db->{hash1}->{subkey2} = "subvalue2";
1213 my $struct = $db->export();
1215 print $struct->{key1} . "\n"; # prints "value1"
1217 This makes a complete copy of the database in memory, and returns a reference
1218 to it. The C<export()> method can be called on any database level (not just
1219 the base level), and works with both hash and array DB types. Be careful of
1220 large databases -- you can store a lot more data in a DBM::Deep object than an
1221 in-memory Perl structure.
1223 B<Note:> Make sure your database has no circular references in it.
1224 These will cause an infinite loop when exporting.
1228 DBM::Deep has a number of hooks where you can specify your own Perl function
1229 to perform filtering on incoming or outgoing data. This is a perfect
1230 way to extend the engine, and implement things like real-time compression or
1231 encryption. Filtering applies to the base DB level, and all child hashes /
1232 arrays. Filter hooks can be specified when your DBM::Deep object is first
1233 constructed, or by calling the C<set_filter()> method at any time. There are
1234 four available filter hooks, described below:
1238 =item * filter_store_key
1240 This filter is called whenever a hash key is stored. It
1241 is passed the incoming key, and expected to return a transformed key.
1243 =item * filter_store_value
1245 This filter is called whenever a hash key or array element is stored. It
1246 is passed the incoming value, and expected to return a transformed value.
1248 =item * filter_fetch_key
1250 This filter is called whenever a hash key is fetched (i.e. via
1251 C<first_key()> or C<next_key()>). It is passed the transformed key,
1252 and expected to return the plain key.
1254 =item * filter_fetch_value
1256 This filter is called whenever a hash key or array element is fetched.
1257 It is passed the transformed value, and expected to return the plain value.
1261 Here are the two ways to setup a filter hook:
1263 my $db = DBM::Deep->new(
1265 filter_store_value => \&my_filter_store,
1266 filter_fetch_value => \&my_filter_fetch
1271 $db->set_filter( "filter_store_value", \&my_filter_store );
1272 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1274 Your filter function will be called only when dealing with SCALAR keys or
1275 values. When nested hashes and arrays are being stored/fetched, filtering
1276 is bypassed. Filters are called as static functions, passed a single SCALAR
1277 argument, and expected to return a single SCALAR value. If you want to
1278 remove a filter, set the function reference to C<undef>:
1280 $db->set_filter( "filter_store_value", undef );
1282 =head2 REAL-TIME ENCRYPTION EXAMPLE
1284 Here is a working example that uses the I<Crypt::Blowfish> module to
1285 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1286 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1287 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1290 use Crypt::Blowfish;
1293 my $cipher = Crypt::CBC->new({
1294 'key' => 'my secret key',
1295 'cipher' => 'Blowfish',
1297 'regenerate_key' => 0,
1298 'padding' => 'space',
1302 my $db = DBM::Deep->new(
1303 file => "foo-encrypt.db",
1304 filter_store_key => \&my_encrypt,
1305 filter_store_value => \&my_encrypt,
1306 filter_fetch_key => \&my_decrypt,
1307 filter_fetch_value => \&my_decrypt,
1310 $db->{key1} = "value1";
1311 $db->{key2} = "value2";
1312 print "key1: " . $db->{key1} . "\n";
1313 print "key2: " . $db->{key2} . "\n";
1319 return $cipher->encrypt( $_[0] );
1322 return $cipher->decrypt( $_[0] );
1325 =head2 REAL-TIME COMPRESSION EXAMPLE
1327 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1328 compression / decompression of keys & values with DBM::Deep Filters.
1329 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1330 more on I<Compress::Zlib>.
1335 my $db = DBM::Deep->new(
1336 file => "foo-compress.db",
1337 filter_store_key => \&my_compress,
1338 filter_store_value => \&my_compress,
1339 filter_fetch_key => \&my_decompress,
1340 filter_fetch_value => \&my_decompress,
1343 $db->{key1} = "value1";
1344 $db->{key2} = "value2";
1345 print "key1: " . $db->{key1} . "\n";
1346 print "key2: " . $db->{key2} . "\n";
1352 return Compress::Zlib::memGzip( $_[0] ) ;
1355 return Compress::Zlib::memGunzip( $_[0] ) ;
1358 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1359 actually numerical index numbers, and are not filtered.
1361 =head1 ERROR HANDLING
1363 Most DBM::Deep methods return a true value for success, and call die() on
1364 failure. You can wrap calls in an eval block to catch the die.
1366 my $db = DBM::Deep->new( "foo.db" ); # create hash
1367 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1369 print $@; # prints error message
1371 =head1 LARGEFILE SUPPORT
1373 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1374 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1375 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1376 by calling the static C<set_pack()> method before you do anything else.
1378 DBM::Deep::set_pack(8, 'Q');
1380 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1381 instead of 32-bit longs. After setting these values your DB files have a
1382 theoretical maximum size of 16 XB (exabytes).
1384 B<Note:> Changing these values will B<NOT> work for existing database files.
1385 Only change this for new files, and make sure it stays set consistently
1386 throughout the file's life. If you do set these values, you can no longer
1387 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
1388 back to 32-bit mode.
1390 B<Note:> I have not personally tested files > 2 GB -- all my systems have
1391 only a 32-bit Perl. However, I have received user reports that this does
1394 =head1 LOW-LEVEL ACCESS
1396 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1397 you can call the C<_fh()> method, which returns the handle:
1399 my $fh = $db->_fh();
1401 This method can be called on the root level of the datbase, or any child
1402 hashes or arrays. All levels share a I<root> structure, which contains things
1403 like the filehandle, a reference counter, and all the options specified
1404 when you created the object. You can get access to this root structure by
1405 calling the C<root()> method.
1407 my $root = $db->_root();
1409 This is useful for changing options after the object has already been created,
1410 such as enabling/disabling locking. You can also store your own temporary user
1411 data in this structure (be wary of name collision), which is then accessible from
1412 any child hash or array.
1414 =head1 CUSTOM DIGEST ALGORITHM
1416 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1417 keys. However you can override this, and use another algorithm (such as SHA-256)
1418 or even write your own. But please note that DBM::Deep currently expects zero
1419 collisions, so your algorithm has to be I<perfect>, so to speak.
1420 Collision detection may be introduced in a later version.
1424 You can specify a custom digest algorithm by calling the static C<set_digest()>
1425 function, passing a reference to a subroutine, and the length of the algorithm's
1426 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
1427 objects. Here is a working example that uses a 256-bit hash from the
1428 I<Digest::SHA256> module. Please see
1429 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
1434 my $context = Digest::SHA256::new(256);
1436 DBM::Deep::set_digest( \&my_digest, 32 );
1438 my $db = DBM::Deep->new( "foo-sha.db" );
1440 $db->{key1} = "value1";
1441 $db->{key2} = "value2";
1442 print "key1: " . $db->{key1} . "\n";
1443 print "key2: " . $db->{key2} . "\n";
1449 return substr( $context->hash($_[0]), 0, 32 );
1452 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1453 of bytes you specify in the C<set_digest()> function (in this case 32).
1455 =head1 CIRCULAR REFERENCES
1457 DBM::Deep has B<experimental> support for circular references. Meaning you
1458 can have a nested hash key or array element that points to a parent object.
1459 This relationship is stored in the DB file, and is preserved between sessions.
1462 my $db = DBM::Deep->new( "foo.db" );
1465 $db->{circle} = $db; # ref to self
1467 print $db->{foo} . "\n"; # prints "bar"
1468 print $db->{circle}->{foo} . "\n"; # prints "bar" again
1470 B<Note>: Passing the object to a function that recursively walks the
1471 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1472 C<export()> methods) will result in an infinite loop. This will be fixed in
1475 =head1 CAVEATS / ISSUES / BUGS
1477 This section describes all the known issues with DBM::Deep. It you have found
1478 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1480 =head2 UNUSED SPACE RECOVERY
1482 One major caveat with DBM::Deep is that space occupied by existing keys and
1483 values is not recovered when they are deleted. Meaning if you keep deleting
1484 and adding new keys, your file will continuously grow. I am working on this,
1485 but in the meantime you can call the built-in C<optimize()> method from time to
1486 time (perhaps in a crontab or something) to recover all your unused space.
1488 $db->optimize(); # returns true on success
1490 This rebuilds the ENTIRE database into a new file, then moves it on top of
1491 the original. The new file will have no unused space, thus it will take up as
1492 little disk space as possible. Please note that this operation can take
1493 a long time for large files, and you need enough disk space to temporarily hold
1494 2 copies of your DB file. The temporary file is created in the same directory
1495 as the original, named with a ".tmp" extension, and is deleted when the
1496 operation completes. Oh, and if locking is enabled, the DB is automatically
1497 locked for the entire duration of the copy.
1499 B<WARNING:> Only call optimize() on the top-level node of the database, and
1500 make sure there are no child references lying around. DBM::Deep keeps a reference
1501 counter, and if it is greater than 1, optimize() will abort and return undef.
1503 =head2 AUTOVIVIFICATION
1505 Unfortunately, autovivification doesn't work with tied hashes. This appears to
1506 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
1507 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
1508 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
1509 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
1512 $db->{foo}->{bar} = "hello";
1514 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
1515 being an empty hash. Try this instead, which works fine:
1517 $db->{foo} = { bar => "hello" };
1519 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
1520 the execution path, and Perl indeed passes an empty hash to the STORE() method.
1521 Probably a bug in Perl.
1525 (The reasons given assume a high level of Perl understanding, specifically of
1526 references. You can safely skip this section.)
1528 Currently, the only references supported are HASH and ARRAY. The other reference
1529 types (SCALAR, CODE, GLOB, and REF) cannot be supported for various reasons.
1535 These are things like filehandles and other sockets. They can't be supported
1536 because it's completely unclear how DBM::Deep should serialize them.
1538 =item * SCALAR / REF
1540 The discussion here refers to the following type of example:
1547 # In some other process ...
1549 my $val = ${ $db->{key1} };
1551 is( $val, 50, "What actually gets stored in the DB file?" );
1553 The problem is one of synchronization. When the variable being referred to
1554 changes value, the reference isn't notified. This means that the new value won't
1555 be stored in the datafile for other processes to read. There is no TIEREF.
1557 It is theoretically possible to store references to values already within a
1558 DBM::Deep object because everything already is synchronized, but the change to
1559 the internals would be quite large. Specifically, DBM::Deep would have to tie
1560 every single value that is stored. This would bloat the RAM footprint of
1561 DBM::Deep at least twofold (if not more) and be a significant performance drain,
1562 all to support a feature that has never been requested.
1566 L<http://search.cpan.org/search?module=Data::Dump::Streamer> provides a
1567 mechanism for serializing coderefs, including saving off all closure state.
1568 However, just as for SCALAR and REF, that closure state may change without
1569 notifying the DBM::Deep object storing the reference.
1573 =head2 FILE CORRUPTION
1575 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1576 for a 32-bit signature when opened, but other corruption in files can cause
1577 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1578 stuck in an infinite loop depending on the level of corruption. File write
1579 operations are not checked for failure (for speed), so if you happen to run
1580 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1581 be addressed in a later version of DBM::Deep.
1585 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1586 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1587 about setting up your NFS server with a locking daemon, then using lockf() to
1588 lock your files, but your mileage may vary there as well. From what I
1589 understand, there is no real way to do it. However, if you need access to the
1590 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1591 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1593 =head2 COPYING OBJECTS
1595 Beware of copying tied objects in Perl. Very strange things can happen.
1596 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1597 returns a new, blessed, tied hash or array to the same level in the DB.
1599 my $copy = $db->clone();
1601 B<Note>: Since clone() here is cloning the object, not the database location, any
1602 modifications to either $db or $copy will be visible in both.
1606 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1607 These functions cause every element in the array to move, which can be murder
1608 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1609 a different location. This will be addressed in the forthcoming version 1.00.
1611 =head2 WRITEONLY FILES
1613 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1614 writeonly mode. STORE will verify that the filehandle is writable. However, there
1615 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1616 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1620 This section discusses DBM::Deep's speed and memory usage.
1624 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1625 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1626 multi-level hash/array support, and cross-platform FTPable files. Even so,
1627 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1628 with huge databases. Here is some test data:
1630 Adding 1,000,000 keys to new DB file...
1632 At 100 keys, avg. speed is 2,703 keys/sec
1633 At 200 keys, avg. speed is 2,642 keys/sec
1634 At 300 keys, avg. speed is 2,598 keys/sec
1635 At 400 keys, avg. speed is 2,578 keys/sec
1636 At 500 keys, avg. speed is 2,722 keys/sec
1637 At 600 keys, avg. speed is 2,628 keys/sec
1638 At 700 keys, avg. speed is 2,700 keys/sec
1639 At 800 keys, avg. speed is 2,607 keys/sec
1640 At 900 keys, avg. speed is 2,190 keys/sec
1641 At 1,000 keys, avg. speed is 2,570 keys/sec
1642 At 2,000 keys, avg. speed is 2,417 keys/sec
1643 At 3,000 keys, avg. speed is 1,982 keys/sec
1644 At 4,000 keys, avg. speed is 1,568 keys/sec
1645 At 5,000 keys, avg. speed is 1,533 keys/sec
1646 At 6,000 keys, avg. speed is 1,787 keys/sec
1647 At 7,000 keys, avg. speed is 1,977 keys/sec
1648 At 8,000 keys, avg. speed is 2,028 keys/sec
1649 At 9,000 keys, avg. speed is 2,077 keys/sec
1650 At 10,000 keys, avg. speed is 2,031 keys/sec
1651 At 20,000 keys, avg. speed is 1,970 keys/sec
1652 At 30,000 keys, avg. speed is 2,050 keys/sec
1653 At 40,000 keys, avg. speed is 2,073 keys/sec
1654 At 50,000 keys, avg. speed is 1,973 keys/sec
1655 At 60,000 keys, avg. speed is 1,914 keys/sec
1656 At 70,000 keys, avg. speed is 2,091 keys/sec
1657 At 80,000 keys, avg. speed is 2,103 keys/sec
1658 At 90,000 keys, avg. speed is 1,886 keys/sec
1659 At 100,000 keys, avg. speed is 1,970 keys/sec
1660 At 200,000 keys, avg. speed is 2,053 keys/sec
1661 At 300,000 keys, avg. speed is 1,697 keys/sec
1662 At 400,000 keys, avg. speed is 1,838 keys/sec
1663 At 500,000 keys, avg. speed is 1,941 keys/sec
1664 At 600,000 keys, avg. speed is 1,930 keys/sec
1665 At 700,000 keys, avg. speed is 1,735 keys/sec
1666 At 800,000 keys, avg. speed is 1,795 keys/sec
1667 At 900,000 keys, avg. speed is 1,221 keys/sec
1668 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1670 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1671 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1672 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1673 Run time was 12 min 3 sec.
1677 One of the great things about DBM::Deep is that it uses very little memory.
1678 Even with huge databases (1,000,000+ keys) you will not see much increased
1679 memory on your process. DBM::Deep relies solely on the filesystem for storing
1680 and fetching data. Here is output from I</usr/bin/top> before even opening a
1683 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1684 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1686 Basically the process is taking 2,716K of memory. And here is the same
1687 process after storing and fetching 1,000,000 keys:
1689 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1690 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1692 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1693 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1695 =head1 DB FILE FORMAT
1697 In case you were interested in the underlying DB file format, it is documented
1698 here in this section. You don't need to know this to use the module, it's just
1699 included for reference.
1703 DBM::Deep files always start with a 32-bit signature to identify the file type.
1704 This is at offset 0. The signature is "DPDB" in network byte order. This is
1705 checked for when the file is opened and an error will be thrown if it's not found.
1709 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
1710 has a standard header containing the type of data, the length of data, and then
1711 the data itself. The type is a single character (1 byte), the length is a
1712 32-bit unsigned long in network byte order, and the data is, well, the data.
1713 Here is how it unfolds:
1717 Immediately after the 32-bit file signature is the I<Master Index> record.
1718 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
1719 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
1720 depending on how the DBM::Deep object was constructed.
1722 The index works by looking at a I<MD5 Hash> of the hash key (or array index
1723 number). The first 8-bit char of the MD5 signature is the offset into the
1724 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
1725 index element is a file offset of the next tag for the key/element in question,
1726 which is usually a I<Bucket List> tag (see below).
1728 The next tag I<could> be another index, depending on how many keys/elements
1729 exist. See L<RE-INDEXING> below for details.
1733 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
1734 file offsets to where the actual data is stored. It starts with a standard
1735 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
1736 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
1737 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
1738 When the list fills up, a I<Re-Index> operation is performed (See
1739 L<RE-INDEXING> below).
1743 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
1744 index/value pair (in array mode). It starts with a standard tag header with
1745 type I<D> for scalar data (string, binary, etc.), or it could be a nested
1746 hash (type I<H>) or array (type I<A>). The value comes just after the tag
1747 header. The size reported in the tag header is only for the value, but then,
1748 just after the value is another size (32-bit unsigned long) and then the plain
1749 key itself. Since the value is likely to be fetched more often than the plain
1750 key, I figured it would be I<slightly> faster to store the value first.
1752 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
1753 record for the nested structure, where the process begins all over again.
1757 After a I<Bucket List> grows to 16 records, its allocated space in the file is
1758 exhausted. Then, when another key/element comes in, the list is converted to a
1759 new index record. However, this index will look at the next char in the MD5
1760 hash, and arrange new Bucket List pointers accordingly. This process is called
1761 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
1762 17 (16 + new one) keys/elements are removed from the old Bucket List and
1763 inserted into the new index. Several new Bucket Lists are created in the
1764 process, as a new MD5 char from the key is being examined (it is unlikely that
1765 the keys will all share the same next char of their MD5s).
1767 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
1768 when the Bucket Lists will turn into indexes, but the first round tends to
1769 happen right around 4,000 keys. You will see a I<slight> decrease in
1770 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
1771 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
1772 right around 900,000 keys. This process can continue nearly indefinitely --
1773 right up until the point the I<MD5> signatures start colliding with each other,
1774 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
1775 getting struck by lightning while you are walking to cash in your tickets.
1776 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
1777 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
1778 this is 340 unodecillion, but don't quote me).
1782 When a new key/element is stored, the key (or index number) is first run through
1783 I<Digest::MD5> to get a 128-bit signature (example, in hex:
1784 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
1785 for the first char of the signature (in this case I<b0>). If it does not exist,
1786 a new I<Bucket List> is created for our key (and the next 15 future keys that
1787 happen to also have I<b> as their first MD5 char). The entire MD5 is written
1788 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
1789 this point, unless we are replacing an existing I<Bucket>), where the actual
1790 data will be stored.
1794 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
1795 (or index number), then walking along the indexes. If there are enough
1796 keys/elements in this DB level, there might be nested indexes, each linked to
1797 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
1798 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
1799 question. If we found a match, the I<Bucket> tag is loaded, where the value and
1800 plain key are stored.
1802 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
1803 methods. In this process the indexes are walked systematically, and each key
1804 fetched in increasing MD5 order (which is why it appears random). Once the
1805 I<Bucket> is found, the value is skipped and the plain key returned instead.
1806 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
1807 alphabetically sorted. This only happens on an index-level -- as soon as the
1808 I<Bucket Lists> are hit, the keys will come out in the order they went in --
1809 so it's pretty much undefined how the keys will come out -- just like Perl's
1812 =head1 CODE COVERAGE
1814 We use B<Devel::Cover> to test the code coverage of our tests, below is the
1815 B<Devel::Cover> report on this module's test suite.
1817 ----------------------------------- ------ ------ ------ ------ ------ ------
1818 File stmt bran cond sub time total
1819 ----------------------------------- ------ ------ ------ ------ ------ ------
1820 blib/lib/DBM/Deep.pm 94.9 80.6 73.0 100.0 37.9 90.4
1821 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 18.2 98.1
1822 blib/lib/DBM/Deep/Engine.pm 98.9 87.3 80.0 100.0 34.2 95.2
1823 blib/lib/DBM/Deep/Hash.pm 100.0 87.5 100.0 100.0 9.7 97.3
1824 Total 97.9 85.9 79.7 100.0 100.0 94.3
1825 ----------------------------------- ------ ------ ------ ------ ------ ------
1827 =head1 MORE INFORMATION
1829 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
1830 or send email to L<DBM-Deep@googlegroups.com>.
1834 Joseph Huckaby, L<jhuckaby@cpan.org>
1836 Rob Kinyon, L<rkinyon@cpan.org>
1838 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
1842 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
1843 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
1847 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
1848 This is free software, you may use it and distribute it under the
1849 same terms as Perl itself.