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 # Set to 4 and 'N' for 32-bit offset tags (default). Theoretical limit of 4 GB per file.
45 # (Perl must be compiled with largefile support for files > 2 GB)
47 # Set to 8 and 'Q' for 64-bit offsets. Theoretical limit of 16 XB per file.
48 # (Perl must be compiled with largefile and 64-bit long support)
54 # Set to 4 and 'N' for 32-bit data length prefixes. Limit of 4 GB for each key/value.
55 # Upgrading this is possible (see above) but probably not necessary. If you need
56 # more than 4 GB for a single key or value, this module is really not for you :-)
58 #my $DATA_LENGTH_SIZE = 4;
59 #my $DATA_LENGTH_PACK = 'N';
60 our ($LONG_SIZE, $LONG_PACK, $DATA_LENGTH_SIZE, $DATA_LENGTH_PACK);
63 # Maximum number of buckets per list before another level of indexing is done.
64 # Increase this value for slightly greater speed, but larger database files.
65 # DO NOT decrease this value below 16, due to risk of recursive reindex overrun.
67 our $MAX_BUCKETS = 16;
70 # Better not adjust anything below here, unless you're me :-)
74 # Setup digest function for keys
76 our ($DIGEST_FUNC, $HASH_SIZE);
77 #my $DIGEST_FUNC = \&Digest::MD5::md5;
80 # Precalculate index and bucket sizes based on values above.
83 our ($INDEX_SIZE, $BUCKET_SIZE, $BUCKET_LIST_SIZE);
90 # Setup file and tag signatures. These should never change.
92 sub SIG_FILE () { 'DPDB' }
93 sub SIG_HASH () { 'H' }
94 sub SIG_ARRAY () { 'A' }
95 sub SIG_SCALAR () { 'S' }
96 sub SIG_NULL () { 'N' }
97 sub SIG_DATA () { 'D' }
98 sub SIG_INDEX () { 'I' }
99 sub SIG_BLIST () { 'B' }
100 sub SIG_SIZE () { 1 }
103 # Setup constants for users to pass to new()
105 sub TYPE_HASH () { SIG_HASH }
106 sub TYPE_ARRAY () { SIG_ARRAY }
107 sub TYPE_SCALAR () { SIG_SCALAR }
113 if (scalar(@_) > 1) {
115 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
119 elsif ( ref $_[0] ) {
120 unless ( eval { local $SIG{'__DIE__'}; %{$_[0]} || 1 } ) {
121 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
126 $args = { file => shift };
134 # Class constructor method for Perl OO interface.
135 # Calls tie() and returns blessed reference to tied hash or array,
136 # providing a hybrid OO/tie interface.
139 my $args = $class->_get_args( @_ );
142 # Check if we want a tied hash or array.
145 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
146 $class = 'DBM::Deep::Array';
147 require DBM::Deep::Array;
148 tie @$self, $class, %$args;
151 $class = 'DBM::Deep::Hash';
152 require DBM::Deep::Hash;
153 tie %$self, $class, %$args;
156 return bless $self, $class;
161 # Setup $self and bless into this class.
166 # These are the defaults to be optionally overridden below
169 base_offset => length(SIG_FILE),
170 engine => 'DBM::Deep::Engine',
173 foreach my $param ( keys %$self ) {
174 next unless exists $args->{$param};
175 $self->{$param} = delete $args->{$param}
178 # locking implicitly enables autoflush
179 if ($args->{locking}) { $args->{autoflush} = 1; }
181 $self->{root} = exists $args->{root}
183 : DBM::Deep::_::Root->new( $args );
185 if (!defined($self->_fh)) { $self->{engine}->open( $self ); }
192 require DBM::Deep::Hash;
193 return DBM::Deep::Hash->TIEHASH( @_ );
198 require DBM::Deep::Array;
199 return DBM::Deep::Array->TIEARRAY( @_ );
202 #XXX Unneeded now ...
208 # Check existence of single key given tag and MD5 digested key.
211 my ($tag, $md5) = @_;
212 my $keys = $tag->{content};
215 # Iterate through buckets, looking for a key match
218 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
219 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
220 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
224 # Hit end of list, no match
229 if ( $md5 ne $key ) {
234 # Matched key -- return true
242 sub _find_bucket_list {
244 # Locate offset for bucket list, given digested key
250 # Locate offset for bucket list using digest index system
253 my $tag = $self->{engine}->load_tag($self, $self->_base_offset);
254 if (!$tag) { return; }
256 while ($tag->{signature} ne SIG_BLIST) {
257 $tag = $self->{engine}->index_lookup($self, $tag, ord(substr($md5, $ch, 1)));
258 if (!$tag) { return; }
265 sub _traverse_index {
267 # Scan index and recursively step into deeper levels, looking for next key.
269 my ($self, $offset, $ch, $force_return_next) = @_;
270 $force_return_next = undef unless $force_return_next;
272 my $tag = $self->{engine}->load_tag($self, $offset );
276 if ($tag->{signature} ne SIG_BLIST) {
277 my $content = $tag->{content};
279 if ($self->{return_next}) { $start = 0; }
280 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
282 for (my $index = $start; $index < 256; $index++) {
283 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
285 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
286 if (defined($result)) { return $result; }
290 $self->{return_next} = 1;
293 elsif ($tag->{signature} eq SIG_BLIST) {
294 my $keys = $tag->{content};
295 if ($force_return_next) { $self->{return_next} = 1; }
298 # Iterate through buckets, looking for a key match
300 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
301 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
302 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
306 # End of bucket list -- return to outer loop
308 $self->{return_next} = 1;
311 elsif ($key eq $self->{prev_md5}) {
313 # Located previous key -- return next one found
315 $self->{return_next} = 1;
318 elsif ($self->{return_next}) {
320 # Seek to bucket location and skip over signature
322 seek($fh, $subloc + SIG_SIZE + $self->_root->{file_offset}, SEEK_SET);
325 # Skip over value to get to plain key
328 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
329 if ($size) { seek($fh, $size, SEEK_CUR); }
332 # Read in plain key and return as scalar
335 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
336 if ($size) { read( $fh, $plain_key, $size); }
342 $self->{return_next} = 1;
343 } # tag is a bucket list
350 # Locate next key, given digested previous one
352 my $self = $_[0]->_get_self;
354 $self->{prev_md5} = $_[1] ? $_[1] : undef;
355 $self->{return_next} = 0;
358 # If the previous key was not specifed, start at the top and
359 # return the first one found.
361 if (!$self->{prev_md5}) {
362 $self->{prev_md5} = chr(0) x $HASH_SIZE;
363 $self->{return_next} = 1;
366 return $self->_traverse_index( $self->_base_offset, 0 );
371 # If db locking is set, flock() the db file. If called multiple
372 # times before unlock(), then the same number of unlocks() must
373 # be called before the lock is released.
375 my $self = $_[0]->_get_self;
377 $type = LOCK_EX unless defined $type;
379 if (!defined($self->_fh)) { return; }
381 if ($self->_root->{locking}) {
382 if (!$self->_root->{locked}) {
383 flock($self->_fh, $type);
385 # refresh end counter in case file has changed size
386 my @stats = stat($self->_root->{file});
387 $self->_root->{end} = $stats[7];
389 # double-check file inode, in case another process
390 # has optimize()d our file while we were waiting.
391 if ($stats[1] != $self->_root->{inode}) {
392 $self->{engine}->open( $self ); # re-open
393 flock($self->_fh, $type); # re-lock
394 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
397 $self->_root->{locked}++;
407 # If db locking is set, unlock the db file. See note in lock()
408 # regarding calling lock() multiple times.
410 my $self = $_[0]->_get_self;
412 if (!defined($self->_fh)) { return; }
414 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
415 $self->_root->{locked}--;
416 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
425 my $self = shift->_get_self;
426 my ($spot, $value) = @_;
431 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
432 my $type = $value->_type;
433 ${$spot} = $type eq TYPE_HASH ? {} : [];
434 $value->_copy_node( ${$spot} );
437 my $r = Scalar::Util::reftype( $value );
438 my $c = Scalar::Util::blessed( $value );
439 if ( $r eq 'ARRAY' ) {
440 ${$spot} = [ @{$value} ];
443 ${$spot} = { %{$value} };
445 ${$spot} = bless ${$spot}, $c
454 # Copy single level of keys or elements to new DB handle.
455 # Recurse for nested structures
457 my $self = shift->_get_self;
460 if ($self->_type eq TYPE_HASH) {
461 my $key = $self->first_key();
463 my $value = $self->get($key);
464 $self->_copy_value( \$db_temp->{$key}, $value );
465 $key = $self->next_key($key);
469 my $length = $self->length();
470 for (my $index = 0; $index < $length; $index++) {
471 my $value = $self->get($index);
472 $self->_copy_value( \$db_temp->[$index], $value );
481 # Recursively export into standard Perl hashes and arrays.
483 my $self = $_[0]->_get_self;
486 if ($self->_type eq TYPE_HASH) { $temp = {}; }
487 elsif ($self->_type eq TYPE_ARRAY) { $temp = []; }
490 $self->_copy_node( $temp );
498 # Recursively import Perl hash/array structure
500 #XXX This use of ref() seems to be ok
501 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
503 my $self = $_[0]->_get_self;
506 #XXX This use of ref() seems to be ok
509 # struct is not a reference, so just import based on our type
513 if ($self->_type eq TYPE_HASH) { $struct = {@_}; }
514 elsif ($self->_type eq TYPE_ARRAY) { $struct = [@_]; }
517 my $r = Scalar::Util::reftype($struct) || '';
518 if ($r eq "HASH" && $self->_type eq TYPE_HASH) {
519 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
521 elsif ($r eq "ARRAY" && $self->_type eq TYPE_ARRAY) {
522 $self->push( @$struct );
525 return $self->_throw_error("Cannot import: type mismatch");
533 # Rebuild entire database into new file, then move
534 # it back on top of original.
536 my $self = $_[0]->_get_self;
538 #XXX Need to create a new test for this
539 # if ($self->_root->{links} > 1) {
540 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
543 my $db_temp = DBM::Deep->new(
544 file => $self->_root->{file} . '.tmp',
548 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
552 $self->_copy_node( $db_temp );
556 # Attempt to copy user, group and permissions over to new file
558 my @stats = stat($self->_fh);
559 my $perms = $stats[2] & 07777;
562 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
563 chmod( $perms, $self->_root->{file} . '.tmp' );
565 # q.v. perlport for more information on this variable
566 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
568 # Potential race condition when optmizing on Win32 with locking.
569 # The Windows filesystem requires that the filehandle be closed
570 # before it is overwritten with rename(). This could be redone
574 $self->{engine}->close( $self );
577 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
578 unlink $self->_root->{file} . '.tmp';
580 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
584 $self->{engine}->close( $self );
585 $self->{engine}->open( $self );
592 # Make copy of object and return
594 my $self = $_[0]->_get_self;
596 return DBM::Deep->new(
597 type => $self->_type,
598 base_offset => $self->_base_offset,
604 my %is_legal_filter = map {
607 store_key store_value
608 fetch_key fetch_value
613 # Setup filter function for storing or fetching the key or value
615 my $self = $_[0]->_get_self;
617 my $func = $_[2] ? $_[2] : undef;
619 if ( $is_legal_filter{$type} ) {
620 $self->_root->{"filter_$type"} = $func;
634 # Get access to the root structure
636 my $self = $_[0]->_get_self;
637 return $self->{root};
642 # Get access to the raw fh
644 #XXX It will be useful, though, when we split out HASH and ARRAY
645 my $self = $_[0]->_get_self;
646 return $self->_root->{fh};
651 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
653 my $self = $_[0]->_get_self;
654 return $self->{type};
659 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
661 my $self = $_[0]->_get_self;
662 return $self->{base_offset};
670 die "DBM::Deep: $_[1]\n";
675 # Precalculate index, bucket and bucket list sizes
678 #XXX I don't like this ...
679 set_pack() unless defined $LONG_SIZE;
681 $INDEX_SIZE = 256 * $LONG_SIZE;
682 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
683 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
688 # Set pack/unpack modes (see file header for more)
690 my ($long_s, $long_p, $data_s, $data_p) = @_;
692 $LONG_SIZE = $long_s ? $long_s : 4;
693 $LONG_PACK = $long_p ? $long_p : 'N';
695 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
696 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
703 # Set key digest function (default is MD5)
705 my ($digest_func, $hash_size) = @_;
707 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
708 $HASH_SIZE = $hash_size ? $hash_size : 16;
715 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
720 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
724 # tie() methods (hashes and arrays)
729 # Store single hash key/value or array element in database.
731 my $self = $_[0]->_get_self;
734 # User may be storing a hash, in which case we do not want it run
735 # through the filtering system
736 my $value = ($self->_root->{filter_store_value} && !ref($_[2]))
737 ? $self->_root->{filter_store_value}->($_[2])
740 my $md5 = $DIGEST_FUNC->($key);
742 unless ( _is_writable( $self->_fh ) ) {
743 $self->_throw_error( 'Cannot write to a readonly filehandle' );
747 # Request exclusive lock for writing
749 $self->lock( LOCK_EX );
754 # Locate offset for bucket list using digest index system
756 my $tag = $self->{engine}->load_tag($self, $self->_base_offset);
758 $tag = $self->{engine}->create_tag($self, $self->_base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
762 while ($tag->{signature} ne SIG_BLIST) {
763 my $num = ord(substr($md5, $ch, 1));
765 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
766 my $new_tag = $self->{engine}->index_lookup($self, $tag, $num);
769 seek($fh, $ref_loc + $self->_root->{file_offset}, SEEK_SET);
770 print( $fh pack($LONG_PACK, $self->_root->{end}) );
772 $tag = $self->{engine}->create_tag($self, $self->_root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
774 $tag->{ref_loc} = $ref_loc;
782 $tag->{ref_loc} = $ref_loc;
789 # Add key/value to bucket list
791 my $result = $self->{engine}->add_bucket( $self, $tag, $md5, $key, $value );
800 # Fetch single value or element given plain key or array index
802 my $self = shift->_get_self;
805 my $md5 = $DIGEST_FUNC->($key);
808 # Request shared lock for reading
810 $self->lock( LOCK_SH );
812 my $tag = $self->_find_bucket_list( $md5 );
819 # Get value from bucket list
821 my $result = $self->{engine}->get_bucket_value( $self, $tag, $md5 );
825 #XXX What is ref() checking here?
826 #YYY Filters only apply on scalar values, so the ref check is making
827 #YYY sure the fetched bucket is a scalar, not a child hash or array.
828 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
829 ? $self->_root->{filter_fetch_value}->($result)
835 # Delete single key/value pair or element given plain key or array index
837 my $self = $_[0]->_get_self;
840 my $md5 = $DIGEST_FUNC->($key);
843 # Request exclusive lock for writing
845 $self->lock( LOCK_EX );
847 my $tag = $self->_find_bucket_list( $md5 );
856 my $value = $self->{engine}->get_bucket_value($self, $tag, $md5 );
857 if ($value && !ref($value) && $self->_root->{filter_fetch_value}) {
858 $value = $self->_root->{filter_fetch_value}->($value);
861 my $result = $self->{engine}->delete_bucket( $self, $tag, $md5 );
864 # If this object is an array and the key deleted was on the end of the stack,
865 # decrement the length variable.
875 # Check if a single key or element exists given plain key or array index
877 my $self = $_[0]->_get_self;
880 my $md5 = $DIGEST_FUNC->($key);
883 # Request shared lock for reading
885 $self->lock( LOCK_SH );
887 my $tag = $self->_find_bucket_list( $md5 );
890 # For some reason, the built-in exists() function returns '' for false
898 # Check if bucket exists and return 1 or ''
900 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
909 # Clear all keys from hash, or all elements from array.
911 my $self = $_[0]->_get_self;
914 # Request exclusive lock for writing
916 $self->lock( LOCK_EX );
920 seek($fh, $self->_base_offset + $self->_root->{file_offset}, SEEK_SET);
926 $self->{engine}->create_tag($self, $self->_base_offset, $self->_type, chr(0) x $INDEX_SIZE);
934 # Public method aliases
936 sub put { (shift)->STORE( @_ ) }
937 sub store { (shift)->STORE( @_ ) }
938 sub get { (shift)->FETCH( @_ ) }
939 sub fetch { (shift)->FETCH( @_ ) }
940 sub delete { (shift)->DELETE( @_ ) }
941 sub exists { (shift)->EXISTS( @_ ) }
942 sub clear { (shift)->CLEAR( @_ ) }
944 package DBM::Deep::_::Root;
958 filter_store_key => undef,
959 filter_store_value => undef,
960 filter_fetch_key => undef,
961 filter_fetch_value => undef,
967 if ( $self->{fh} && !$self->{file_offset} ) {
968 $self->{file_offset} = tell( $self->{fh} );
978 close $self->{fh} if $self->{fh};
989 DBM::Deep - A pure perl multi-level hash/array DBM
994 my $db = DBM::Deep->new( "foo.db" );
996 $db->{key} = 'value'; # tie() style
999 $db->put('key' => 'value'); # OO style
1000 print $db->get('key');
1002 # true multi-level support
1003 $db->{my_complex} = [
1004 'hello', { perl => 'rules' },
1010 A unique flat-file database module, written in pure perl. True
1011 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1012 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1013 handle millions of keys and unlimited hash levels without significant
1014 slow-down. Written from the ground-up in pure perl -- this is NOT a
1015 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1016 Mac OS X and Windows.
1020 Hopefully you are using Perl's excellent CPAN module, which will download
1021 and install the module for you. If not, get the tarball, and run these
1033 Construction can be done OO-style (which is the recommended way), or using
1034 Perl's tie() function. Both are examined here.
1036 =head2 OO CONSTRUCTION
1038 The recommended way to construct a DBM::Deep object is to use the new()
1039 method, which gets you a blessed, tied hash or array reference.
1041 my $db = DBM::Deep->new( "foo.db" );
1043 This opens a new database handle, mapped to the file "foo.db". If this
1044 file does not exist, it will automatically be created. DB files are
1045 opened in "r+" (read/write) mode, and the type of object returned is a
1046 hash, unless otherwise specified (see L<OPTIONS> below).
1048 You can pass a number of options to the constructor to specify things like
1049 locking, autoflush, etc. This is done by passing an inline hash:
1051 my $db = DBM::Deep->new(
1057 Notice that the filename is now specified I<inside> the hash with
1058 the "file" parameter, as opposed to being the sole argument to the
1059 constructor. This is required if any options are specified.
1060 See L<OPTIONS> below for the complete list.
1064 You can also start with an array instead of a hash. For this, you must
1065 specify the C<type> parameter:
1067 my $db = DBM::Deep->new(
1069 type => DBM::Deep->TYPE_ARRAY
1072 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1073 a new DB file. If you create a DBM::Deep object with an existing file, the
1074 C<type> will be loaded from the file header, and an error will be thrown if
1075 the wrong type is passed in.
1077 =head2 TIE CONSTRUCTION
1079 Alternately, you can create a DBM::Deep handle by using Perl's built-in
1080 tie() function. The object returned from tie() can be used to call methods,
1081 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
1082 file (as expected with most tie'd objects).
1085 my $db = tie %hash, "DBM::Deep", "foo.db";
1088 my $db = tie @array, "DBM::Deep", "bar.db";
1090 As with the OO constructor, you can replace the DB filename parameter with
1091 a hash containing one or more options (see L<OPTIONS> just below for the
1094 tie %hash, "DBM::Deep", {
1102 There are a number of options that can be passed in when constructing your
1103 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1109 Filename of the DB file to link the handle to. You can pass a full absolute
1110 filesystem path, partial path, or a plain filename if the file is in the
1111 current working directory. This is a required parameter (though q.v. fh).
1115 If you want, you can pass in the fh instead of the file. This is most useful for doing
1118 my $db = DBM::Deep->new( { fh => \*DATA } );
1120 You are responsible for making sure that the fh has been opened appropriately for your
1121 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
1122 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
1123 needs to read from the fh.
1127 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
1128 not need to set this. However, it's there if you want it.
1130 If you pass in fh and do not set this, it will be set appropriately.
1134 This parameter specifies what type of object to create, a hash or array. Use
1135 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1136 This only takes effect when beginning a new file. This is an optional
1137 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1141 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1142 function to lock the database in exclusive mode for writes, and shared mode for
1143 reads. Pass any true value to enable. This affects the base DB handle I<and
1144 any child hashes or arrays> that use the same DB file. This is an optional
1145 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1149 Specifies whether autoflush is to be enabled on the underlying filehandle.
1150 This obviously slows down write operations, but is required if you may have
1151 multiple processes accessing the same DB file (also consider enable I<locking>).
1152 Pass any true value to enable. This is an optional parameter, and defaults to 0
1157 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1158 restore them when fetched. This is an B<experimental> feature, and does have
1159 side-effects. Basically, when hashes are re-blessed into their original
1160 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1161 able to call any DBM::Deep methods on them. You have been warned.
1162 This is an optional parameter, and defaults to 0 (disabled).
1166 See L<FILTERS> below.
1170 Setting I<debug> mode will make all errors non-fatal, dump them out to
1171 STDERR, and continue on. This is for debugging purposes only, and probably
1172 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1174 B<NOTE>: This parameter is considered deprecated and should not be used anymore.
1178 =head1 TIE INTERFACE
1180 With DBM::Deep you can access your databases using Perl's standard hash/array
1181 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
1182 treat them as such. DBM::Deep will intercept all reads/writes and direct them
1183 to the right place -- the DB file. This has nothing to do with the
1184 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
1185 using regular hashes and arrays, rather than calling functions like C<get()>
1186 and C<put()> (although those work too). It is entirely up to you how to want
1187 to access your databases.
1191 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1192 or even nested hashes (or arrays) using standard Perl syntax:
1194 my $db = DBM::Deep->new( "foo.db" );
1196 $db->{mykey} = "myvalue";
1198 $db->{myhash}->{subkey} = "subvalue";
1200 print $db->{myhash}->{subkey} . "\n";
1202 You can even step through hash keys using the normal Perl C<keys()> function:
1204 foreach my $key (keys %$db) {
1205 print "$key: " . $db->{$key} . "\n";
1208 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1209 pushes them onto an array, all before the loop even begins. If you have an
1210 extra large hash, this may exhaust Perl's memory. Instead, consider using
1211 Perl's C<each()> function, which pulls keys/values one at a time, using very
1214 while (my ($key, $value) = each %$db) {
1215 print "$key: $value\n";
1218 Please note that when using C<each()>, you should always pass a direct
1219 hash reference, not a lookup. Meaning, you should B<never> do this:
1222 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1224 This causes an infinite loop, because for each iteration, Perl is calling
1225 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1226 it effectively keeps returning the first key over and over again. Instead,
1227 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1231 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1232 reference. This includes inserting, removing and manipulating elements,
1233 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1234 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1235 or simply be a nested array reference inside a hash. Example:
1237 my $db = DBM::Deep->new(
1238 file => "foo-array.db",
1239 type => DBM::Deep->TYPE_ARRAY
1243 push @$db, "bar", "baz";
1244 unshift @$db, "bah";
1246 my $last_elem = pop @$db; # baz
1247 my $first_elem = shift @$db; # bah
1248 my $second_elem = $db->[1]; # bar
1250 my $num_elements = scalar @$db;
1254 In addition to the I<tie()> interface, you can also use a standard OO interface
1255 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1256 array) has its own methods, but both types share the following common methods:
1257 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1261 =item * new() / clone()
1263 These are the constructor and copy-functions.
1265 =item * put() / store()
1267 Stores a new hash key/value pair, or sets an array element value. Takes two
1268 arguments, the hash key or array index, and the new value. The value can be
1269 a scalar, hash ref or array ref. Returns true on success, false on failure.
1271 $db->put("foo", "bar"); # for hashes
1272 $db->put(1, "bar"); # for arrays
1274 =item * get() / fetch()
1276 Fetches the value of a hash key or array element. Takes one argument: the hash
1277 key or array index. Returns a scalar, hash ref or array ref, depending on the
1280 my $value = $db->get("foo"); # for hashes
1281 my $value = $db->get(1); # for arrays
1285 Checks if a hash key or array index exists. Takes one argument: the hash key
1286 or array index. Returns true if it exists, false if not.
1288 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1289 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1293 Deletes one hash key/value pair or array element. Takes one argument: the hash
1294 key or array index. Returns true on success, false if not found. For arrays,
1295 the remaining elements located after the deleted element are NOT moved over.
1296 The deleted element is essentially just undefined, which is exactly how Perl's
1297 internal arrays work. Please note that the space occupied by the deleted
1298 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1299 below for details and workarounds.
1301 $db->delete("foo"); # for hashes
1302 $db->delete(1); # for arrays
1306 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1307 value. Please note that the space occupied by the deleted keys/values or
1308 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1309 details and workarounds.
1311 $db->clear(); # hashes or arrays
1313 =item * lock() / unlock()
1319 Recover lost disk space.
1321 =item * import() / export()
1323 Data going in and out.
1325 =item * set_digest() / set_pack() / set_filter()
1327 q.v. adjusting the interal parameters.
1333 For hashes, DBM::Deep supports all the common methods described above, and the
1334 following additional methods: C<first_key()> and C<next_key()>.
1340 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1341 fetched in an undefined order (which appears random). Takes no arguments,
1342 returns the key as a scalar value.
1344 my $key = $db->first_key();
1348 Returns the "next" key in the hash, given the previous one as the sole argument.
1349 Returns undef if there are no more keys to be fetched.
1351 $key = $db->next_key($key);
1355 Here are some examples of using hashes:
1357 my $db = DBM::Deep->new( "foo.db" );
1359 $db->put("foo", "bar");
1360 print "foo: " . $db->get("foo") . "\n";
1362 $db->put("baz", {}); # new child hash ref
1363 $db->get("baz")->put("buz", "biz");
1364 print "buz: " . $db->get("baz")->get("buz") . "\n";
1366 my $key = $db->first_key();
1368 print "$key: " . $db->get($key) . "\n";
1369 $key = $db->next_key($key);
1372 if ($db->exists("foo")) { $db->delete("foo"); }
1376 For arrays, DBM::Deep supports all the common methods described above, and the
1377 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1378 C<unshift()> and C<splice()>.
1384 Returns the number of elements in the array. Takes no arguments.
1386 my $len = $db->length();
1390 Adds one or more elements onto the end of the array. Accepts scalars, hash
1391 refs or array refs. No return value.
1393 $db->push("foo", "bar", {});
1397 Fetches the last element in the array, and deletes it. Takes no arguments.
1398 Returns undef if array is empty. Returns the element value.
1400 my $elem = $db->pop();
1404 Fetches the first element in the array, deletes it, then shifts all the
1405 remaining elements over to take up the space. Returns the element value. This
1406 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1409 my $elem = $db->shift();
1413 Inserts one or more elements onto the beginning of the array, shifting all
1414 existing elements over to make room. Accepts scalars, hash refs or array refs.
1415 No return value. This method is not recommended with large arrays -- see
1416 <LARGE ARRAYS> below for details.
1418 $db->unshift("foo", "bar", {});
1422 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1423 -f splice> for usage -- it is too complicated to document here. This method is
1424 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1428 Here are some examples of using arrays:
1430 my $db = DBM::Deep->new(
1432 type => DBM::Deep->TYPE_ARRAY
1435 $db->push("bar", "baz");
1436 $db->unshift("foo");
1439 my $len = $db->length();
1440 print "length: $len\n"; # 4
1442 for (my $k=0; $k<$len; $k++) {
1443 print "$k: " . $db->get($k) . "\n";
1446 $db->splice(1, 2, "biz", "baf");
1448 while (my $elem = shift @$db) {
1449 print "shifted: $elem\n";
1454 Enable automatic file locking by passing a true value to the C<locking>
1455 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1457 my $db = DBM::Deep->new(
1462 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1463 mode for writes, and shared mode for reads. This is required if you have
1464 multiple processes accessing the same database file, to avoid file corruption.
1465 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1466 NFS> below for more.
1468 =head2 EXPLICIT LOCKING
1470 You can explicitly lock a database, so it remains locked for multiple
1471 transactions. This is done by calling the C<lock()> method, and passing an
1472 optional lock mode argument (defaults to exclusive mode). This is particularly
1473 useful for things like counters, where the current value needs to be fetched,
1474 then incremented, then stored again.
1477 my $counter = $db->get("counter");
1479 $db->put("counter", $counter);
1488 You can pass C<lock()> an optional argument, which specifies which mode to use
1489 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1490 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1491 same as the constants defined in Perl's C<Fcntl> module.
1493 $db->lock( DBM::Deep->LOCK_SH );
1497 =head1 IMPORTING/EXPORTING
1499 You can import existing complex structures by calling the C<import()> method,
1500 and export an entire database into an in-memory structure using the C<export()>
1501 method. Both are examined here.
1505 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1506 walking the structure and adding keys/elements to the database as you go,
1507 simply pass a reference to the C<import()> method. This recursively adds
1508 everything to an existing DBM::Deep object for you. Here is an example:
1513 array1 => [ "elem0", "elem1", "elem2" ],
1515 subkey1 => "subvalue1",
1516 subkey2 => "subvalue2"
1520 my $db = DBM::Deep->new( "foo.db" );
1521 $db->import( $struct );
1523 print $db->{key1} . "\n"; # prints "value1"
1525 This recursively imports the entire C<$struct> object into C<$db>, including
1526 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1527 keys are merged with the existing ones, replacing if they already exist.
1528 The C<import()> method can be called on any database level (not just the base
1529 level), and works with both hash and array DB types.
1531 B<Note:> Make sure your existing structure has no circular references in it.
1532 These will cause an infinite loop when importing.
1536 Calling the C<export()> method on an existing DBM::Deep object will return
1537 a reference to a new in-memory copy of the database. The export is done
1538 recursively, so all nested hashes/arrays are all exported to standard Perl
1539 objects. Here is an example:
1541 my $db = DBM::Deep->new( "foo.db" );
1543 $db->{key1} = "value1";
1544 $db->{key2} = "value2";
1546 $db->{hash1}->{subkey1} = "subvalue1";
1547 $db->{hash1}->{subkey2} = "subvalue2";
1549 my $struct = $db->export();
1551 print $struct->{key1} . "\n"; # prints "value1"
1553 This makes a complete copy of the database in memory, and returns a reference
1554 to it. The C<export()> method can be called on any database level (not just
1555 the base level), and works with both hash and array DB types. Be careful of
1556 large databases -- you can store a lot more data in a DBM::Deep object than an
1557 in-memory Perl structure.
1559 B<Note:> Make sure your database has no circular references in it.
1560 These will cause an infinite loop when exporting.
1564 DBM::Deep has a number of hooks where you can specify your own Perl function
1565 to perform filtering on incoming or outgoing data. This is a perfect
1566 way to extend the engine, and implement things like real-time compression or
1567 encryption. Filtering applies to the base DB level, and all child hashes /
1568 arrays. Filter hooks can be specified when your DBM::Deep object is first
1569 constructed, or by calling the C<set_filter()> method at any time. There are
1570 four available filter hooks, described below:
1574 =item * filter_store_key
1576 This filter is called whenever a hash key is stored. It
1577 is passed the incoming key, and expected to return a transformed key.
1579 =item * filter_store_value
1581 This filter is called whenever a hash key or array element is stored. It
1582 is passed the incoming value, and expected to return a transformed value.
1584 =item * filter_fetch_key
1586 This filter is called whenever a hash key is fetched (i.e. via
1587 C<first_key()> or C<next_key()>). It is passed the transformed key,
1588 and expected to return the plain key.
1590 =item * filter_fetch_value
1592 This filter is called whenever a hash key or array element is fetched.
1593 It is passed the transformed value, and expected to return the plain value.
1597 Here are the two ways to setup a filter hook:
1599 my $db = DBM::Deep->new(
1601 filter_store_value => \&my_filter_store,
1602 filter_fetch_value => \&my_filter_fetch
1607 $db->set_filter( "filter_store_value", \&my_filter_store );
1608 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1610 Your filter function will be called only when dealing with SCALAR keys or
1611 values. When nested hashes and arrays are being stored/fetched, filtering
1612 is bypassed. Filters are called as static functions, passed a single SCALAR
1613 argument, and expected to return a single SCALAR value. If you want to
1614 remove a filter, set the function reference to C<undef>:
1616 $db->set_filter( "filter_store_value", undef );
1618 =head2 REAL-TIME ENCRYPTION EXAMPLE
1620 Here is a working example that uses the I<Crypt::Blowfish> module to
1621 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1622 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1623 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1626 use Crypt::Blowfish;
1629 my $cipher = Crypt::CBC->new({
1630 'key' => 'my secret key',
1631 'cipher' => 'Blowfish',
1633 'regenerate_key' => 0,
1634 'padding' => 'space',
1638 my $db = DBM::Deep->new(
1639 file => "foo-encrypt.db",
1640 filter_store_key => \&my_encrypt,
1641 filter_store_value => \&my_encrypt,
1642 filter_fetch_key => \&my_decrypt,
1643 filter_fetch_value => \&my_decrypt,
1646 $db->{key1} = "value1";
1647 $db->{key2} = "value2";
1648 print "key1: " . $db->{key1} . "\n";
1649 print "key2: " . $db->{key2} . "\n";
1655 return $cipher->encrypt( $_[0] );
1658 return $cipher->decrypt( $_[0] );
1661 =head2 REAL-TIME COMPRESSION EXAMPLE
1663 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1664 compression / decompression of keys & values with DBM::Deep Filters.
1665 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1666 more on I<Compress::Zlib>.
1671 my $db = DBM::Deep->new(
1672 file => "foo-compress.db",
1673 filter_store_key => \&my_compress,
1674 filter_store_value => \&my_compress,
1675 filter_fetch_key => \&my_decompress,
1676 filter_fetch_value => \&my_decompress,
1679 $db->{key1} = "value1";
1680 $db->{key2} = "value2";
1681 print "key1: " . $db->{key1} . "\n";
1682 print "key2: " . $db->{key2} . "\n";
1688 return Compress::Zlib::memGzip( $_[0] ) ;
1691 return Compress::Zlib::memGunzip( $_[0] ) ;
1694 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1695 actually numerical index numbers, and are not filtered.
1697 =head1 ERROR HANDLING
1699 Most DBM::Deep methods return a true value for success, and call die() on
1700 failure. You can wrap calls in an eval block to catch the die.
1702 my $db = DBM::Deep->new( "foo.db" ); # create hash
1703 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1705 print $@; # prints error message
1707 =head1 LARGEFILE SUPPORT
1709 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1710 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1711 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1712 by calling the static C<set_pack()> method before you do anything else.
1714 DBM::Deep::set_pack(8, 'Q');
1716 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1717 instead of 32-bit longs. After setting these values your DB files have a
1718 theoretical maximum size of 16 XB (exabytes).
1720 B<Note:> Changing these values will B<NOT> work for existing database files.
1721 Only change this for new files, and make sure it stays set consistently
1722 throughout the file's life. If you do set these values, you can no longer
1723 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
1724 back to 32-bit mode.
1726 B<Note:> I have not personally tested files > 2 GB -- all my systems have
1727 only a 32-bit Perl. However, I have received user reports that this does
1730 =head1 LOW-LEVEL ACCESS
1732 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1733 you can call the C<_fh()> method, which returns the handle:
1735 my $fh = $db->_fh();
1737 This method can be called on the root level of the datbase, or any child
1738 hashes or arrays. All levels share a I<root> structure, which contains things
1739 like the filehandle, a reference counter, and all the options specified
1740 when you created the object. You can get access to this root structure by
1741 calling the C<root()> method.
1743 my $root = $db->_root();
1745 This is useful for changing options after the object has already been created,
1746 such as enabling/disabling locking, or debug modes. You can also
1747 store your own temporary user data in this structure (be wary of name
1748 collision), which is then accessible from any child hash or array.
1750 =head1 CUSTOM DIGEST ALGORITHM
1752 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1753 keys. However you can override this, and use another algorithm (such as SHA-256)
1754 or even write your own. But please note that DBM::Deep currently expects zero
1755 collisions, so your algorithm has to be I<perfect>, so to speak.
1756 Collision detection may be introduced in a later version.
1760 You can specify a custom digest algorithm by calling the static C<set_digest()>
1761 function, passing a reference to a subroutine, and the length of the algorithm's
1762 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
1763 objects. Here is a working example that uses a 256-bit hash from the
1764 I<Digest::SHA256> module. Please see
1765 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
1770 my $context = Digest::SHA256::new(256);
1772 DBM::Deep::set_digest( \&my_digest, 32 );
1774 my $db = DBM::Deep->new( "foo-sha.db" );
1776 $db->{key1} = "value1";
1777 $db->{key2} = "value2";
1778 print "key1: " . $db->{key1} . "\n";
1779 print "key2: " . $db->{key2} . "\n";
1785 return substr( $context->hash($_[0]), 0, 32 );
1788 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1789 of bytes you specify in the C<set_digest()> function (in this case 32).
1791 =head1 CIRCULAR REFERENCES
1793 DBM::Deep has B<experimental> support for circular references. Meaning you
1794 can have a nested hash key or array element that points to a parent object.
1795 This relationship is stored in the DB file, and is preserved between sessions.
1798 my $db = DBM::Deep->new( "foo.db" );
1801 $db->{circle} = $db; # ref to self
1803 print $db->{foo} . "\n"; # prints "foo"
1804 print $db->{circle}->{foo} . "\n"; # prints "foo" again
1806 One catch is, passing the object to a function that recursively walks the
1807 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1808 C<export()> methods) will result in an infinite loop. The other catch is,
1809 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
1810 or C<next_key()> methods), you will get the I<target object's key>, not the
1811 ref's key. This gets even more interesting with the above example, where
1812 the I<circle> key points to the base DB object, which technically doesn't
1813 have a key. So I made DBM::Deep return "[base]" as the key name in that
1816 =head1 CAVEATS / ISSUES / BUGS
1818 This section describes all the known issues with DBM::Deep. It you have found
1819 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1821 =head2 UNUSED SPACE RECOVERY
1823 One major caveat with DBM::Deep is that space occupied by existing keys and
1824 values is not recovered when they are deleted. Meaning if you keep deleting
1825 and adding new keys, your file will continuously grow. I am working on this,
1826 but in the meantime you can call the built-in C<optimize()> method from time to
1827 time (perhaps in a crontab or something) to recover all your unused space.
1829 $db->optimize(); # returns true on success
1831 This rebuilds the ENTIRE database into a new file, then moves it on top of
1832 the original. The new file will have no unused space, thus it will take up as
1833 little disk space as possible. Please note that this operation can take
1834 a long time for large files, and you need enough disk space to temporarily hold
1835 2 copies of your DB file. The temporary file is created in the same directory
1836 as the original, named with a ".tmp" extension, and is deleted when the
1837 operation completes. Oh, and if locking is enabled, the DB is automatically
1838 locked for the entire duration of the copy.
1840 B<WARNING:> Only call optimize() on the top-level node of the database, and
1841 make sure there are no child references lying around. DBM::Deep keeps a reference
1842 counter, and if it is greater than 1, optimize() will abort and return undef.
1844 =head2 AUTOVIVIFICATION
1846 Unfortunately, autovivification doesn't work with tied hashes. This appears to
1847 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
1848 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
1849 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
1850 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
1853 $db->{foo}->{bar} = "hello";
1855 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
1856 being an empty hash. Try this instead, which works fine:
1858 $db->{foo} = { bar => "hello" };
1860 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
1861 the execution path, and Perl indeed passes an empty hash to the STORE() method.
1862 Probably a bug in Perl.
1864 =head2 FILE CORRUPTION
1866 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1867 for a 32-bit signature when opened, but other corruption in files can cause
1868 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1869 stuck in an infinite loop depending on the level of corruption. File write
1870 operations are not checked for failure (for speed), so if you happen to run
1871 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1872 be addressed in a later version of DBM::Deep.
1876 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1877 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1878 about setting up your NFS server with a locking daemon, then using lockf() to
1879 lock your files, but your mileage may vary there as well. From what I
1880 understand, there is no real way to do it. However, if you need access to the
1881 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1882 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1884 =head2 COPYING OBJECTS
1886 Beware of copying tied objects in Perl. Very strange things can happen.
1887 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1888 returns a new, blessed, tied hash or array to the same level in the DB.
1890 my $copy = $db->clone();
1892 B<Note>: Since clone() here is cloning the object, not the database location, any
1893 modifications to either $db or $copy will be visible in both.
1897 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1898 These functions cause every element in the array to move, which can be murder
1899 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1900 a different location. This will be addressed in the forthcoming version 1.00.
1902 =head2 WRITEONLY FILES
1904 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1905 writeonly mode. STORE will verify that the filehandle is writable. However, there
1906 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1907 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1911 This section discusses DBM::Deep's speed and memory usage.
1915 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1916 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1917 multi-level hash/array support, and cross-platform FTPable files. Even so,
1918 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1919 with huge databases. Here is some test data:
1921 Adding 1,000,000 keys to new DB file...
1923 At 100 keys, avg. speed is 2,703 keys/sec
1924 At 200 keys, avg. speed is 2,642 keys/sec
1925 At 300 keys, avg. speed is 2,598 keys/sec
1926 At 400 keys, avg. speed is 2,578 keys/sec
1927 At 500 keys, avg. speed is 2,722 keys/sec
1928 At 600 keys, avg. speed is 2,628 keys/sec
1929 At 700 keys, avg. speed is 2,700 keys/sec
1930 At 800 keys, avg. speed is 2,607 keys/sec
1931 At 900 keys, avg. speed is 2,190 keys/sec
1932 At 1,000 keys, avg. speed is 2,570 keys/sec
1933 At 2,000 keys, avg. speed is 2,417 keys/sec
1934 At 3,000 keys, avg. speed is 1,982 keys/sec
1935 At 4,000 keys, avg. speed is 1,568 keys/sec
1936 At 5,000 keys, avg. speed is 1,533 keys/sec
1937 At 6,000 keys, avg. speed is 1,787 keys/sec
1938 At 7,000 keys, avg. speed is 1,977 keys/sec
1939 At 8,000 keys, avg. speed is 2,028 keys/sec
1940 At 9,000 keys, avg. speed is 2,077 keys/sec
1941 At 10,000 keys, avg. speed is 2,031 keys/sec
1942 At 20,000 keys, avg. speed is 1,970 keys/sec
1943 At 30,000 keys, avg. speed is 2,050 keys/sec
1944 At 40,000 keys, avg. speed is 2,073 keys/sec
1945 At 50,000 keys, avg. speed is 1,973 keys/sec
1946 At 60,000 keys, avg. speed is 1,914 keys/sec
1947 At 70,000 keys, avg. speed is 2,091 keys/sec
1948 At 80,000 keys, avg. speed is 2,103 keys/sec
1949 At 90,000 keys, avg. speed is 1,886 keys/sec
1950 At 100,000 keys, avg. speed is 1,970 keys/sec
1951 At 200,000 keys, avg. speed is 2,053 keys/sec
1952 At 300,000 keys, avg. speed is 1,697 keys/sec
1953 At 400,000 keys, avg. speed is 1,838 keys/sec
1954 At 500,000 keys, avg. speed is 1,941 keys/sec
1955 At 600,000 keys, avg. speed is 1,930 keys/sec
1956 At 700,000 keys, avg. speed is 1,735 keys/sec
1957 At 800,000 keys, avg. speed is 1,795 keys/sec
1958 At 900,000 keys, avg. speed is 1,221 keys/sec
1959 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1961 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1962 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1963 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1964 Run time was 12 min 3 sec.
1968 One of the great things about DBM::Deep is that it uses very little memory.
1969 Even with huge databases (1,000,000+ keys) you will not see much increased
1970 memory on your process. DBM::Deep relies solely on the filesystem for storing
1971 and fetching data. Here is output from I</usr/bin/top> before even opening a
1974 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1975 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1977 Basically the process is taking 2,716K of memory. And here is the same
1978 process after storing and fetching 1,000,000 keys:
1980 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1981 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1983 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1984 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1986 =head1 DB FILE FORMAT
1988 In case you were interested in the underlying DB file format, it is documented
1989 here in this section. You don't need to know this to use the module, it's just
1990 included for reference.
1994 DBM::Deep files always start with a 32-bit signature to identify the file type.
1995 This is at offset 0. The signature is "DPDB" in network byte order. This is
1996 checked for when the file is opened and an error will be thrown if it's not found.
2000 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2001 has a standard header containing the type of data, the length of data, and then
2002 the data itself. The type is a single character (1 byte), the length is a
2003 32-bit unsigned long in network byte order, and the data is, well, the data.
2004 Here is how it unfolds:
2008 Immediately after the 32-bit file signature is the I<Master Index> record.
2009 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2010 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2011 depending on how the DBM::Deep object was constructed.
2013 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2014 number). The first 8-bit char of the MD5 signature is the offset into the
2015 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2016 index element is a file offset of the next tag for the key/element in question,
2017 which is usually a I<Bucket List> tag (see below).
2019 The next tag I<could> be another index, depending on how many keys/elements
2020 exist. See L<RE-INDEXING> below for details.
2024 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2025 file offsets to where the actual data is stored. It starts with a standard
2026 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2027 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2028 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2029 When the list fills up, a I<Re-Index> operation is performed (See
2030 L<RE-INDEXING> below).
2034 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2035 index/value pair (in array mode). It starts with a standard tag header with
2036 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2037 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2038 header. The size reported in the tag header is only for the value, but then,
2039 just after the value is another size (32-bit unsigned long) and then the plain
2040 key itself. Since the value is likely to be fetched more often than the plain
2041 key, I figured it would be I<slightly> faster to store the value first.
2043 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2044 record for the nested structure, where the process begins all over again.
2048 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2049 exhausted. Then, when another key/element comes in, the list is converted to a
2050 new index record. However, this index will look at the next char in the MD5
2051 hash, and arrange new Bucket List pointers accordingly. This process is called
2052 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2053 17 (16 + new one) keys/elements are removed from the old Bucket List and
2054 inserted into the new index. Several new Bucket Lists are created in the
2055 process, as a new MD5 char from the key is being examined (it is unlikely that
2056 the keys will all share the same next char of their MD5s).
2058 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2059 when the Bucket Lists will turn into indexes, but the first round tends to
2060 happen right around 4,000 keys. You will see a I<slight> decrease in
2061 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2062 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2063 right around 900,000 keys. This process can continue nearly indefinitely --
2064 right up until the point the I<MD5> signatures start colliding with each other,
2065 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2066 getting struck by lightning while you are walking to cash in your tickets.
2067 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2068 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2069 this is 340 unodecillion, but don't quote me).
2073 When a new key/element is stored, the key (or index number) is first run through
2074 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2075 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2076 for the first char of the signature (in this case I<b0>). If it does not exist,
2077 a new I<Bucket List> is created for our key (and the next 15 future keys that
2078 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2079 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2080 this point, unless we are replacing an existing I<Bucket>), where the actual
2081 data will be stored.
2085 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2086 (or index number), then walking along the indexes. If there are enough
2087 keys/elements in this DB level, there might be nested indexes, each linked to
2088 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2089 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2090 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2091 plain key are stored.
2093 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2094 methods. In this process the indexes are walked systematically, and each key
2095 fetched in increasing MD5 order (which is why it appears random). Once the
2096 I<Bucket> is found, the value is skipped and the plain key returned instead.
2097 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2098 alphabetically sorted. This only happens on an index-level -- as soon as the
2099 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2100 so it's pretty much undefined how the keys will come out -- just like Perl's
2103 =head1 CODE COVERAGE
2105 We use B<Devel::Cover> to test the code coverage of our tests, below is the
2106 B<Devel::Cover> report on this module's test suite.
2108 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2109 File stmt bran cond sub pod time total
2110 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2111 blib/lib/DBM/Deep.pm 95.2 83.8 70.0 98.2 100.0 58.0 91.0
2112 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 n/a 26.7 98.0
2113 blib/lib/DBM/Deep/Hash.pm 95.3 80.0 100.0 100.0 n/a 15.3 92.4
2114 Total 96.2 84.8 74.4 98.8 100.0 100.0 92.4
2115 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2117 =head1 MORE INFORMATION
2119 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
2120 or send email to L<DBM-Deep@googlegroups.com>.
2124 Joseph Huckaby, L<jhuckaby@cpan.org>
2126 Rob Kinyon, L<rkinyon@cpan.org>
2128 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2132 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2133 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2137 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
2138 This is free software, you may use it and distribute it under the
2139 same terms as Perl itself.