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 " . $db->{my_complex}->[1]->{perl} . "!\n";
27 # (c) 2002-2005 Joseph Huckaby. All Rights Reserved.
28 # This program is free software; you can redistribute it and/or
29 # modify it under the same terms as Perl itself.
38 use vars qw/$VERSION/;
43 # Set to 4 and 'N' for 32-bit offset tags (default). Theoretical limit of 4 GB per file.
44 # (Perl must be compiled with largefile support for files > 2 GB)
46 # Set to 8 and 'Q' for 64-bit offsets. Theoretical limit of 16 XB per file.
47 # (Perl must be compiled with largefile and 64-bit long support)
53 # Set to 4 and 'N' for 32-bit data length prefixes. Limit of 4 GB for each key/value.
54 # Upgrading this is possible (see above) but probably not necessary. If you need
55 # more than 4 GB for a single key or value, this module is really not for you :-)
57 #my $DATA_LENGTH_SIZE = 4;
58 #my $DATA_LENGTH_PACK = 'N';
59 my ($LONG_SIZE, $LONG_PACK, $DATA_LENGTH_SIZE, $DATA_LENGTH_PACK);
62 # Maximum number of buckets per list before another level of indexing is done.
63 # Increase this value for slightly greater speed, but larger database files.
64 # DO NOT decrease this value below 16, due to risk of recursive reindex overrun.
69 # Better not adjust anything below here, unless you're me :-)
73 # Setup digest function for keys
75 my ($DIGEST_FUNC, $HASH_SIZE);
76 #my $DIGEST_FUNC = \&Digest::MD5::md5;
79 # Precalculate index and bucket sizes based on values above.
82 my ($INDEX_SIZE, $BUCKET_SIZE, $BUCKET_LIST_SIZE);
89 # Setup file and tag signatures. These should never change.
91 sub SIG_FILE () { 'DPDB' }
92 sub SIG_HASH () { 'H' }
93 sub SIG_ARRAY () { 'A' }
94 sub SIG_NULL () { 'N' }
95 sub SIG_DATA () { 'D' }
96 sub SIG_INDEX () { 'I' }
97 sub SIG_BLIST () { 'B' }
101 # Setup constants for users to pass to new()
103 sub TYPE_HASH () { return SIG_HASH; }
104 sub TYPE_ARRAY () { return SIG_ARRAY; }
108 # Class constructor method for Perl OO interface.
109 # Calls tie() and returns blessed reference to tied hash or array,
110 # providing a hybrid OO/tie interface.
114 if (scalar(@_) > 1) { $args = {@_}; }
115 else { $args = { file => shift }; }
118 # Check if we want a tied hash or array.
121 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
122 tie @$self, $class, %$args;
125 tie %$self, $class, %$args;
128 return bless $self, $class;
132 my @outer_params = qw( type base_offset );
135 # Setup $self and bless into this class.
142 base_offset => length(SIG_FILE),
153 filter_store_key => undef,
154 filter_store_value => undef,
155 filter_fetch_key => undef,
156 filter_fetch_value => undef,
165 foreach my $outer_parm ( @outer_params ) {
166 next unless exists $args->{$outer_parm};
167 $self->{$outer_parm} = $args->{$outer_parm}
170 if ( exists $args->{root} ) {
171 $self->{root} = $args->{root};
174 # This is cleanup based on the fact that the $args
175 # coming in is for both the root and non-root items
176 delete $self->root->{$_} for @outer_params;
178 $self->root->{links}++;
180 if (!defined($self->fh)) { $self->_open(); }
186 sub _get_self { tied( %{$_[0]} ) || $_[0] }
190 # Tied hash constructor method, called by Perl's tie() function.
194 if (scalar(@_) > 1) { $args = {@_}; }
195 #XXX This use of ref() is bad and is a bug
196 elsif (ref($_[0])) { $args = $_[0]; }
197 else { $args = { file => shift }; }
199 $args->{type} = TYPE_HASH;
201 return $class->_init($args);
206 # Tied array constructor method, called by Perl's tie() function.
210 if (scalar(@_) > 1) { $args = {@_}; }
211 #XXX This use of ref() is bad and is a bug
212 elsif (ref($_[0])) { $args = $_[0]; }
213 else { $args = { file => shift }; }
215 $args->{type} = TYPE_ARRAY;
217 return $class->_init($args);
222 # Class deconstructor. Close file handle if there are no more refs.
224 my $self = _get_self($_[0]);
227 $self->root->{links}--;
229 if (!$self->root->{links}) {
236 # Open a FileHandle to the database, create if nonexistent.
237 # Make sure file signature matches DeepDB spec.
239 my $self = _get_self($_[0]);
241 if (defined($self->fh)) { $self->_close(); }
244 if (!(-e $self->root->{file}) && $self->root->{mode} eq 'r+') {
245 my $temp = FileHandle->new( $self->root->{file}, 'w' );
248 #XXX Convert to set_fh()
249 $self->root->{fh} = FileHandle->new( $self->root->{file}, $self->root->{mode} );
250 # }; if ($@ ) { $self->_throw_error( "Received error: $@\n" ); }
251 if (! defined($self->fh)) {
252 return $self->_throw_error("Cannot open file: " . $self->root->{file} . ": $!");
255 binmode $self->fh; # for win32
256 if ($self->root->{autoflush}) {
257 $self->fh->autoflush();
261 seek($self->fh, 0, 0);
262 my $bytes_read = read( $self->fh, $signature, length(SIG_FILE));
265 # File is empty -- write signature and master index
268 seek($self->fh, 0, 0);
269 $self->fh->print(SIG_FILE);
270 $self->root->{end} = length(SIG_FILE);
271 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
273 my $plain_key = "[base]";
274 $self->fh->print( pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
275 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
282 # Check signature was valid
284 unless ($signature eq SIG_FILE) {
286 return $self->_throw_error("Signature not found -- file is not a Deep DB");
289 $self->root->{end} = (stat($self->fh))[7];
292 # Get our type from master index signature
294 my $tag = $self->_load_tag($self->base_offset);
295 #XXX We probably also want to store the hash algorithm name and not assume anything
297 return $self->_throw_error("Corrupted file, no master index record");
299 if ($self->{type} ne $tag->{signature}) {
300 return $self->_throw_error("File type mismatch");
308 # Close database FileHandle
310 my $self = _get_self($_[0]);
311 undef $self->root->{fh};
316 # Given offset, signature and content, create tag and write to disk
318 my ($self, $offset, $sig, $content) = @_;
319 my $size = length($content);
321 seek($self->fh, $offset, 0);
322 $self->fh->print( $sig . pack($DATA_LENGTH_PACK, $size) . $content );
324 if ($offset == $self->root->{end}) {
325 $self->root->{end} += SIG_SIZE + $DATA_LENGTH_SIZE + $size;
331 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
338 # Given offset, load single tag and return signature, size and data
345 seek($fh, $offset, 0);
346 if (eof $fh) { return undef; }
349 read( $fh, $sig, SIG_SIZE);
352 read( $fh, $size, $DATA_LENGTH_SIZE);
353 $size = unpack($DATA_LENGTH_PACK, $size);
356 read( $fh, $buffer, $size);
361 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
368 # Given index tag, lookup single entry in index and return .
371 my ($tag, $index) = @_;
373 my $location = unpack($LONG_PACK, substr($tag->{content}, $index * $LONG_SIZE, $LONG_SIZE) );
374 if (!$location) { return; }
376 return $self->_load_tag( $location );
381 # Adds one key/value pair to bucket list, given offset, MD5 digest of key,
382 # plain (undigested) key and value.
385 my ($tag, $md5, $plain_key, $value) = @_;
386 my $keys = $tag->{content};
390 my $is_dbm_deep = eval { $value->isa( 'DBM::Deep' ) };
391 my $internal_ref = $is_dbm_deep && ($value->root eq $self->root);
394 # Iterate through buckets, seeing if this is a new entry or a replace.
396 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
397 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
398 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
401 # Found empty bucket (end of list). Populate and exit loop.
405 $location = $internal_ref
406 ? $value->base_offset
407 : $self->root->{end};
409 seek($self->fh, $tag->{offset} + ($i * $BUCKET_SIZE), 0);
410 $self->fh->print( $md5 . pack($LONG_PACK, $location) );
413 elsif ($md5 eq $key) {
415 # Found existing bucket with same key. Replace with new value.
420 $location = $value->base_offset;
421 seek($self->fh, $tag->{offset} + ($i * $BUCKET_SIZE), 0);
422 $self->fh->print( $md5 . pack($LONG_PACK, $location) );
425 seek($self->fh, $subloc + SIG_SIZE, 0);
427 read( $self->fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
430 # If value is a hash, array, or raw value with equal or less size, we can
431 # reuse the same content area of the database. Otherwise, we have to create
432 # a new content area at the EOF.
435 my $r = Scalar::Util::reftype( $value ) || '';
436 if ( $r eq 'HASH' || $r eq 'ARRAY' ) { $actual_length = $INDEX_SIZE; }
437 else { $actual_length = length($value); }
439 if ($actual_length <= $size) {
443 $location = $self->root->{end};
444 seek($self->fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE, 0);
445 $self->fh->print( pack($LONG_PACK, $location) );
453 # If this is an internal reference, return now.
454 # No need to write value or plain key
461 # If bucket didn't fit into list, split into a new index level
464 seek($self->fh, $tag->{ref_loc}, 0);
465 $self->fh->print( pack($LONG_PACK, $self->root->{end}) );
467 my $index_tag = $self->_create_tag($self->root->{end}, SIG_INDEX, chr(0) x $INDEX_SIZE);
470 $keys .= $md5 . pack($LONG_PACK, 0);
472 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
473 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
475 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
476 my $num = ord(substr($key, $tag->{ch} + 1, 1));
478 if ($offsets[$num]) {
479 my $offset = $offsets[$num] + SIG_SIZE + $DATA_LENGTH_SIZE;
480 seek($self->fh, $offset, 0);
482 read( $self->fh, $subkeys, $BUCKET_LIST_SIZE);
484 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
485 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
487 seek($self->fh, $offset + ($k * $BUCKET_SIZE), 0);
488 $self->fh->print( $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
494 $offsets[$num] = $self->root->{end};
495 seek($self->fh, $index_tag->{offset} + ($num * $LONG_SIZE), 0);
496 $self->fh->print( pack($LONG_PACK, $self->root->{end}) );
498 my $blist_tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
500 seek($self->fh, $blist_tag->{offset}, 0);
501 $self->fh->print( $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
506 $location ||= $self->root->{end};
507 } # re-index bucket list
510 # Seek to content area and store signature, value and plaintext key
514 seek($self->fh, $location, 0);
517 # Write signature based on content type, set content length and write actual value.
519 my $r = Scalar::Util::reftype($value) || '';
521 $self->fh->print( TYPE_HASH );
522 $self->fh->print( pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
523 $content_length = $INDEX_SIZE;
525 elsif ($r eq 'ARRAY') {
526 $self->fh->print( TYPE_ARRAY );
527 $self->fh->print( pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
528 $content_length = $INDEX_SIZE;
530 elsif (!defined($value)) {
531 $self->fh->print( SIG_NULL );
532 $self->fh->print( pack($DATA_LENGTH_PACK, 0) );
536 $self->fh->print( SIG_DATA );
537 $self->fh->print( pack($DATA_LENGTH_PACK, length($value)) . $value );
538 $content_length = length($value);
542 # Plain key is stored AFTER value, as keys are typically fetched less often.
544 $self->fh->print( pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
547 # If value is blessed, preserve class name
549 if ( $self->root->{autobless} ) {
550 my $value_class = Scalar::Util::blessed($value);
551 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
553 # Blessed ref -- will restore later
555 $self->fh->print( chr(1) );
556 $self->fh->print( pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
557 $content_length += 1;
558 $content_length += $DATA_LENGTH_SIZE + length($value_class);
561 $self->fh->print( chr(0) );
562 $content_length += 1;
567 # If this is a new content area, advance EOF counter
569 if ($location == $self->root->{end}) {
570 $self->root->{end} += SIG_SIZE;
571 $self->root->{end} += $DATA_LENGTH_SIZE + $content_length;
572 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
576 # If content is a hash or array, create new child DeepDB object and
577 # pass each key or element to it.
580 my $branch = DBM::Deep->new(
582 base_offset => $location,
585 foreach my $key (keys %{$value}) {
586 $branch->{$key} = $value->{$key};
589 elsif ($r eq 'ARRAY') {
590 my $branch = DBM::Deep->new(
592 base_offset => $location,
596 foreach my $element (@{$value}) {
597 $branch->[$index] = $element;
605 return $self->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
608 sub _get_bucket_value {
610 # Fetch single value given tag and MD5 digested key.
613 my ($tag, $md5) = @_;
614 my $keys = $tag->{content};
617 # Iterate through buckets, looking for a key match
620 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
621 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
622 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
626 # Hit end of list, no match
631 if ( $md5 ne $key ) {
636 # Found match -- seek to offset and read signature
639 seek($self->fh, $subloc, 0);
640 read( $self->fh, $signature, SIG_SIZE);
643 # If value is a hash or array, return new DeepDB object with correct offset
645 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
646 my $obj = DBM::Deep->new(
648 base_offset => $subloc,
652 if ($self->root->{autobless}) {
654 # Skip over value and plain key to see if object needs
657 seek($self->fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, 1);
660 read( $self->fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
661 if ($size) { seek($self->fh, $size, 1); }
664 read( $self->fh, $bless_bit, 1);
665 if (ord($bless_bit)) {
667 # Yes, object needs to be re-blessed
670 read( $self->fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
671 if ($size) { read( $self->fh, $class_name, $size); }
672 if ($class_name) { $obj = bless( $obj, $class_name ); }
680 # Otherwise return actual value
682 elsif ($signature eq SIG_DATA) {
685 read( $self->fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
686 if ($size) { read( $self->fh, $value, $size); }
691 # Key exists, but content is null
701 # Delete single key/value pair given tag and MD5 digested key.
704 my ($tag, $md5) = @_;
705 my $keys = $tag->{content};
708 # Iterate through buckets, looking for a key match
711 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
712 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
713 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
717 # Hit end of list, no match
722 if ( $md5 ne $key ) {
727 # Matched key -- delete bucket and return
729 seek($self->fh, $tag->{offset} + ($i * $BUCKET_SIZE), 0);
730 $self->fh->print( substr($keys, ($i+1) * $BUCKET_SIZE ) );
731 $self->fh->print( chr(0) x $BUCKET_SIZE );
741 # Check existence of single key given tag and MD5 digested key.
744 my ($tag, $md5) = @_;
745 my $keys = $tag->{content};
748 # Iterate through buckets, looking for a key match
751 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
752 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
753 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
757 # Hit end of list, no match
762 if ( $md5 ne $key ) {
767 # Matched key -- return true
775 sub _find_bucket_list {
777 # Locate offset for bucket list, given digested key
783 # Locate offset for bucket list using digest index system
786 my $tag = $self->_load_tag($self->base_offset);
787 if (!$tag) { return; }
789 while ($tag->{signature} ne SIG_BLIST) {
790 $tag = $self->_index_lookup($tag, ord(substr($md5, $ch, 1)));
791 if (!$tag) { return; }
798 sub _traverse_index {
800 # Scan index and recursively step into deeper levels, looking for next key.
802 my ($self, $offset, $ch, $force_return_next) = @_;
803 $force_return_next = undef unless $force_return_next;
805 my $tag = $self->_load_tag( $offset );
807 if ($tag->{signature} ne SIG_BLIST) {
808 my $content = $tag->{content};
810 if ($self->{return_next}) { $start = 0; }
811 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
813 for (my $index = $start; $index < 256; $index++) {
814 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
816 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
817 if (defined($result)) { return $result; }
821 $self->{return_next} = 1;
824 elsif ($tag->{signature} eq SIG_BLIST) {
825 my $keys = $tag->{content};
826 if ($force_return_next) { $self->{return_next} = 1; }
829 # Iterate through buckets, looking for a key match
831 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
832 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
833 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
837 # End of bucket list -- return to outer loop
839 $self->{return_next} = 1;
842 elsif ($key eq $self->{prev_md5}) {
844 # Located previous key -- return next one found
846 $self->{return_next} = 1;
849 elsif ($self->{return_next}) {
851 # Seek to bucket location and skip over signature
853 seek($self->fh, $subloc + SIG_SIZE, 0);
856 # Skip over value to get to plain key
859 read( $self->fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
860 if ($size) { seek($self->fh, $size, 1); }
863 # Read in plain key and return as scalar
866 read( $self->fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
867 if ($size) { read( $self->fh, $plain_key, $size); }
873 $self->{return_next} = 1;
874 } # tag is a bucket list
881 # Locate next key, given digested previous one
883 my $self = _get_self($_[0]);
885 $self->{prev_md5} = $_[1] ? $_[1] : undef;
886 $self->{return_next} = 0;
889 # If the previous key was not specifed, start at the top and
890 # return the first one found.
892 if (!$self->{prev_md5}) {
893 $self->{prev_md5} = chr(0) x $HASH_SIZE;
894 $self->{return_next} = 1;
897 return $self->_traverse_index( $self->base_offset, 0 );
902 # If db locking is set, flock() the db file. If called multiple
903 # times before unlock(), then the same number of unlocks() must
904 # be called before the lock is released.
906 my $self = _get_self($_[0]);
908 $type = LOCK_EX unless defined $type;
910 if ($self->root->{locking}) {
911 if (!$self->root->{locked}) { flock($self->fh, $type); }
912 $self->root->{locked}++;
918 # If db locking is set, unlock the db file. See note in lock()
919 # regarding calling lock() multiple times.
921 my $self = _get_self($_[0]);
923 if ($self->root->{locking} && $self->root->{locked} > 0) {
924 $self->root->{locked}--;
925 if (!$self->root->{locked}) { flock($self->fh, LOCK_UN); }
929 #XXX These uses of ref() need verified
932 # Copy single level of keys or elements to new DB handle.
933 # Recurse for nested structures
935 my $self = _get_self($_[0]);
938 if ($self->type eq TYPE_HASH) {
939 my $key = $self->first_key();
941 my $value = $self->get($key);
942 #XXX This doesn't work with autobless
943 if (!ref($value)) { $db_temp->{$key} = $value; }
945 my $type = $value->type;
946 if ($type eq TYPE_HASH) { $db_temp->{$key} = {}; }
947 else { $db_temp->{$key} = []; }
948 $value->_copy_node( $db_temp->{$key} );
950 $key = $self->next_key($key);
954 my $length = $self->length();
955 for (my $index = 0; $index < $length; $index++) {
956 my $value = $self->get($index);
957 if (!ref($value)) { $db_temp->[$index] = $value; }
958 #XXX NO tests for this code
960 my $type = $value->type;
961 if ($type eq TYPE_HASH) { $db_temp->[$index] = {}; }
962 else { $db_temp->[$index] = []; }
963 $value->_copy_node( $db_temp->[$index] );
971 # Recursively export into standard Perl hashes and arrays.
973 my $self = _get_self($_[0]);
976 if ($self->type eq TYPE_HASH) { $temp = {}; }
977 elsif ($self->type eq TYPE_ARRAY) { $temp = []; }
980 $self->_copy_node( $temp );
988 # Recursively import Perl hash/array structure
990 #XXX This use of ref() seems to be ok
991 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
993 my $self = _get_self($_[0]);
996 #XXX This use of ref() seems to be ok
999 # struct is not a reference, so just import based on our type
1003 if ($self->type eq TYPE_HASH) { $struct = {@_}; }
1004 elsif ($self->type eq TYPE_ARRAY) { $struct = [@_]; }
1007 my $r = Scalar::Util::reftype($struct) || '';
1008 if ($r eq "HASH" && $self->type eq TYPE_HASH) {
1009 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
1011 elsif ($r eq "ARRAY" && $self->type eq TYPE_ARRAY) {
1012 $self->push( @$struct );
1015 return $self->_throw_error("Cannot import: type mismatch");
1023 # Rebuild entire database into new file, then move
1024 # it back on top of original.
1026 my $self = _get_self($_[0]);
1027 if ($self->root->{links} > 1) {
1028 return $self->_throw_error("Cannot optimize: reference count is greater than 1");
1031 my $db_temp = DBM::Deep->new(
1032 file => $self->root->{file} . '.tmp',
1036 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
1040 $self->_copy_node( $db_temp );
1044 # Attempt to copy user, group and permissions over to new file
1046 my @stats = stat($self->fh);
1047 my $perms = $stats[2] & 07777;
1048 my $uid = $stats[4];
1049 my $gid = $stats[5];
1050 chown( $uid, $gid, $self->root->{file} . '.tmp' );
1051 chmod( $perms, $self->root->{file} . '.tmp' );
1053 # q.v. perlport for more information on this variable
1054 if ( $^O eq 'MSWin32' ) {
1056 # Potential race condition when optmizing on Win32 with locking.
1057 # The Windows filesystem requires that the filehandle be closed
1058 # before it is overwritten with rename(). This could be redone
1065 if (!rename $self->root->{file} . '.tmp', $self->root->{file}) {
1066 unlink $self->root->{file} . '.tmp';
1068 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
1080 # Make copy of object and return
1082 my $self = _get_self($_[0]);
1084 return DBM::Deep->new(
1085 type => $self->type,
1086 base_offset => $self->base_offset,
1092 my %is_legal_filter = map {
1095 store_key store_value
1096 fetch_key fetch_value
1101 # Setup filter function for storing or fetching the key or value
1103 my $self = _get_self($_[0]);
1104 my $type = lc $_[1];
1105 my $func = $_[2] ? $_[2] : undef;
1107 if ( $is_legal_filter{$type} ) {
1108 $self->root->{"filter_$type"} = $func;
1122 # Get access to the root structure
1124 my $self = _get_self($_[0]);
1125 return $self->{root};
1130 # Get access to the raw FileHandle
1132 #XXX It will be useful, though, when we split out HASH and ARRAY
1133 my $self = _get_self($_[0]);
1134 return $self->root->{fh};
1139 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
1141 my $self = _get_self($_[0]);
1142 return $self->{type};
1147 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
1149 my $self = _get_self($_[0]);
1150 return $self->{base_offset};
1155 # Get last error string, or undef if no error
1158 ? ( _get_self($_[0])->{root}->{error} or undef )
1168 # Store error string in self
1170 my $self = _get_self($_[0]);
1171 my $error_text = $_[1];
1173 $self->root->{error} = $error_text;
1175 unless ($self->root->{debug}) {
1176 die "DBM::Deep: $error_text\n";
1179 warn "DBM::Deep: $error_text\n";
1187 my $self = _get_self($_[0]);
1189 undef $self->root->{error};
1194 # Precalculate index, bucket and bucket list sizes
1197 #XXX I don't like this ...
1198 set_pack() unless defined $LONG_SIZE;
1200 $INDEX_SIZE = 256 * $LONG_SIZE;
1201 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
1202 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
1207 # Set pack/unpack modes (see file header for more)
1209 my ($long_s, $long_p, $data_s, $data_p) = @_;
1211 $LONG_SIZE = $long_s ? $long_s : 4;
1212 $LONG_PACK = $long_p ? $long_p : 'N';
1214 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
1215 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
1222 # Set key digest function (default is MD5)
1224 my ($digest_func, $hash_size) = @_;
1226 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
1227 $HASH_SIZE = $hash_size ? $hash_size : 16;
1233 # tie() methods (hashes and arrays)
1238 # Store single hash key/value or array element in database.
1240 my $self = _get_self($_[0]);
1241 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1242 #XXX What is ref() checking here?
1243 #YYY User may be storing a hash, in which case we do not want it run
1244 #YYY through the filtering system
1245 my $value = ($self->root->{filter_store_value} && !ref($_[2])) ? $self->root->{filter_store_value}->($_[2]) : $_[2];
1247 my $unpacked_key = $key;
1248 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1249 my $md5 = $DIGEST_FUNC->($key);
1252 # Make sure file is open
1254 if (!defined($self->fh) && !$self->_open()) {
1259 # Request exclusive lock for writing
1261 $self->lock( LOCK_EX );
1264 # If locking is enabled, set 'end' parameter again, in case another
1265 # DB instance appended to our file while we were unlocked.
1267 if ($self->root->{locking} || $self->root->{volatile}) {
1268 $self->root->{end} = (stat($self->fh))[7];
1272 # Locate offset for bucket list using digest index system
1274 my $tag = $self->_load_tag($self->base_offset);
1276 $tag = $self->_create_tag($self->base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1280 while ($tag->{signature} ne SIG_BLIST) {
1281 my $num = ord(substr($md5, $ch, 1));
1282 my $new_tag = $self->_index_lookup($tag, $num);
1284 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1285 seek($self->fh, $ref_loc, 0);
1286 $self->fh->print( pack($LONG_PACK, $self->root->{end}) );
1288 $tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1289 $tag->{ref_loc} = $ref_loc;
1294 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1296 $tag->{ref_loc} = $ref_loc;
1303 # Add key/value to bucket list
1305 my $result = $self->_add_bucket( $tag, $md5, $key, $value );
1308 # If this object is an array, and bucket was not a replace, and key is numerical,
1309 # and index is equal or greater than current length, advance length variable.
1311 if (($result == 2) && ($self->type eq TYPE_ARRAY) && ($unpacked_key =~ /^\d+$/) && ($unpacked_key >= $self->FETCHSIZE())) {
1312 $self->STORESIZE( $unpacked_key + 1 );
1322 # Fetch single value or element given plain key or array index
1324 my $self = _get_self($_[0]);
1327 if ( $self->type eq TYPE_HASH ) {
1328 if ( my $filter = $self->root->{filter_store_key} ) {
1329 $key = $filter->( $key );
1332 elsif ( $self->type eq TYPE_ARRAY ) {
1333 if ( $key =~ /^\d+$/ ) {
1334 $key = pack($LONG_PACK, $key);
1338 my $md5 = $DIGEST_FUNC->($key);
1341 # Make sure file is open
1343 if (!defined($self->fh)) { $self->_open(); }
1346 # Request shared lock for reading
1348 $self->lock( LOCK_SH );
1350 my $tag = $self->_find_bucket_list( $md5 );
1357 # Get value from bucket list
1359 my $result = $self->_get_bucket_value( $tag, $md5 );
1363 #XXX What is ref() checking here?
1364 return ($result && !ref($result) && $self->root->{filter_fetch_value}) ? $self->root->{filter_fetch_value}->($result) : $result;
1369 # Delete single key/value pair or element given plain key or array index
1371 my $self = _get_self($_[0]);
1372 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1374 my $unpacked_key = $key;
1375 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1376 my $md5 = $DIGEST_FUNC->($key);
1379 # Make sure file is open
1381 if (!defined($self->fh)) { $self->_open(); }
1384 # Request exclusive lock for writing
1386 $self->lock( LOCK_EX );
1388 my $tag = $self->_find_bucket_list( $md5 );
1397 my $result = $self->_delete_bucket( $tag, $md5 );
1400 # If this object is an array and the key deleted was on the end of the stack,
1401 # decrement the length variable.
1403 if ($result && ($self->type eq TYPE_ARRAY) && ($unpacked_key == $self->FETCHSIZE() - 1)) {
1404 $self->STORESIZE( $unpacked_key );
1414 # Check if a single key or element exists given plain key or array index
1416 my $self = _get_self($_[0]);
1417 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1419 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1420 my $md5 = $DIGEST_FUNC->($key);
1423 # Make sure file is open
1425 if (!defined($self->fh)) { $self->_open(); }
1428 # Request shared lock for reading
1430 $self->lock( LOCK_SH );
1432 my $tag = $self->_find_bucket_list( $md5 );
1435 # For some reason, the built-in exists() function returns '' for false
1443 # Check if bucket exists and return 1 or ''
1445 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1454 # Clear all keys from hash, or all elements from array.
1456 my $self = _get_self($_[0]);
1459 # Make sure file is open
1461 if (!defined($self->fh)) { $self->_open(); }
1464 # Request exclusive lock for writing
1466 $self->lock( LOCK_EX );
1470 seek($fh, $self->base_offset, 0);
1476 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
1485 # Locate and return first key (in no particular order)
1487 my $self = _get_self($_[0]);
1488 if ($self->type ne TYPE_HASH) {
1489 return $self->_throw_error("FIRSTKEY method only supported for hashes");
1493 # Make sure file is open
1495 if (!defined($self->fh)) { $self->_open(); }
1498 # Request shared lock for reading
1500 $self->lock( LOCK_SH );
1502 my $result = $self->_get_next_key();
1506 return ($result && $self->root->{filter_fetch_key}) ? $self->root->{filter_fetch_key}->($result) : $result;
1511 # Return next key (in no particular order), given previous one
1513 my $self = _get_self($_[0]);
1514 if ($self->type ne TYPE_HASH) {
1515 return $self->_throw_error("NEXTKEY method only supported for hashes");
1517 my $prev_key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1518 my $prev_md5 = $DIGEST_FUNC->($prev_key);
1521 # Make sure file is open
1523 if (!defined($self->fh)) { $self->_open(); }
1526 # Request shared lock for reading
1528 $self->lock( LOCK_SH );
1530 my $result = $self->_get_next_key( $prev_md5 );
1534 return ($result && $self->root->{filter_fetch_key}) ? $self->root->{filter_fetch_key}->($result) : $result;
1538 # The following methods are for arrays only
1543 # Return the length of the array
1545 my $self = _get_self($_[0]);
1546 if ($self->type ne TYPE_ARRAY) {
1547 return $self->_throw_error("FETCHSIZE method only supported for arrays");
1550 my $SAVE_FILTER = $self->root->{filter_fetch_value};
1551 $self->root->{filter_fetch_value} = undef;
1553 my $packed_size = $self->FETCH('length');
1555 $self->root->{filter_fetch_value} = $SAVE_FILTER;
1557 if ($packed_size) { return int(unpack($LONG_PACK, $packed_size)); }
1563 # Set the length of the array
1565 my $self = _get_self($_[0]);
1566 if ($self->type ne TYPE_ARRAY) {
1567 return $self->_throw_error("STORESIZE method only supported for arrays");
1569 my $new_length = $_[1];
1571 my $SAVE_FILTER = $self->root->{filter_store_value};
1572 $self->root->{filter_store_value} = undef;
1574 my $result = $self->STORE('length', pack($LONG_PACK, $new_length));
1576 $self->root->{filter_store_value} = $SAVE_FILTER;
1583 # Remove and return the last element on the array
1585 my $self = _get_self($_[0]);
1586 if ($self->type ne TYPE_ARRAY) {
1587 return $self->_throw_error("POP method only supported for arrays");
1589 my $length = $self->FETCHSIZE();
1592 my $content = $self->FETCH( $length - 1 );
1593 $self->DELETE( $length - 1 );
1603 # Add new element(s) to the end of the array
1605 my $self = _get_self(shift);
1606 if ($self->type ne TYPE_ARRAY) {
1607 return $self->_throw_error("PUSH method only supported for arrays");
1609 my $length = $self->FETCHSIZE();
1611 while (my $content = shift @_) {
1612 $self->STORE( $length, $content );
1619 # Remove and return first element on the array.
1620 # Shift over remaining elements to take up space.
1622 my $self = _get_self($_[0]);
1623 if ($self->type ne TYPE_ARRAY) {
1624 return $self->_throw_error("SHIFT method only supported for arrays");
1626 my $length = $self->FETCHSIZE();
1629 my $content = $self->FETCH( 0 );
1632 # Shift elements over and remove last one.
1634 for (my $i = 0; $i < $length - 1; $i++) {
1635 $self->STORE( $i, $self->FETCH($i + 1) );
1637 $self->DELETE( $length - 1 );
1648 # Insert new element(s) at beginning of array.
1649 # Shift over other elements to make space.
1651 my $self = _get_self($_[0]);shift @_;
1652 if ($self->type ne TYPE_ARRAY) {
1653 return $self->_throw_error("UNSHIFT method only supported for arrays");
1655 my @new_elements = @_;
1656 my $length = $self->FETCHSIZE();
1657 my $new_size = scalar @new_elements;
1660 for (my $i = $length - 1; $i >= 0; $i--) {
1661 $self->STORE( $i + $new_size, $self->FETCH($i) );
1665 for (my $i = 0; $i < $new_size; $i++) {
1666 $self->STORE( $i, $new_elements[$i] );
1672 # Splices section of array with optional new section.
1673 # Returns deleted section, or last element deleted in scalar context.
1675 my $self = _get_self($_[0]);shift @_;
1676 if ($self->type ne TYPE_ARRAY) {
1677 return $self->_throw_error("SPLICE method only supported for arrays");
1679 my $length = $self->FETCHSIZE();
1682 # Calculate offset and length of splice
1684 my $offset = shift || 0;
1685 if ($offset < 0) { $offset += $length; }
1688 if (scalar @_) { $splice_length = shift; }
1689 else { $splice_length = $length - $offset; }
1690 if ($splice_length < 0) { $splice_length += ($length - $offset); }
1693 # Setup array with new elements, and copy out old elements for return
1695 my @new_elements = @_;
1696 my $new_size = scalar @new_elements;
1698 my @old_elements = ();
1699 for (my $i = $offset; $i < $offset + $splice_length; $i++) {
1700 push @old_elements, $self->FETCH( $i );
1704 # Adjust array length, and shift elements to accomodate new section.
1706 if ( $new_size != $splice_length ) {
1707 if ($new_size > $splice_length) {
1708 for (my $i = $length - 1; $i >= $offset + $splice_length; $i--) {
1709 $self->STORE( $i + ($new_size - $splice_length), $self->FETCH($i) );
1713 for (my $i = $offset + $splice_length; $i < $length; $i++) {
1714 $self->STORE( $i + ($new_size - $splice_length), $self->FETCH($i) );
1716 for (my $i = 0; $i < $splice_length - $new_size; $i++) {
1717 $self->DELETE( $length - 1 );
1724 # Insert new elements into array
1726 for (my $i = $offset; $i < $offset + $new_size; $i++) {
1727 $self->STORE( $i, shift @new_elements );
1731 # Return deleted section, or last element in scalar context.
1733 return wantarray ? @old_elements : $old_elements[-1];
1736 #XXX We don't need to define it.
1737 #XXX It will be useful, though, when we split out HASH and ARRAY
1740 # Perl will call EXTEND() when the array is likely to grow.
1741 # We don't care, but include it for compatibility.
1746 # Public method aliases
1748 *put = *store = *STORE;
1749 *get = *fetch = *FETCH;
1753 *first_key = *FIRSTKEY;
1754 *next_key = *NEXTKEY;
1755 *length = *FETCHSIZE;
1759 *unshift = *UNSHIFT;
1768 DBM::Deep - A pure perl multi-level hash/array DBM
1773 my $db = DBM::Deep->new( "foo.db" );
1775 $db->{key} = 'value'; # tie() style
1778 $db->put('key', 'value'); # OO style
1779 print $db->get('key');
1781 # true multi-level support
1782 $db->{my_complex} = [
1783 'hello', { perl => 'rules' },
1788 A unique flat-file database module, written in pure perl. True
1789 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1790 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1791 handle millions of keys and unlimited hash levels without significant
1792 slow-down. Written from the ground-up in pure perl -- this is NOT a
1793 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1794 Mac OS X and Windows.
1798 Hopefully you are using CPAN's excellent Perl module, which will download
1799 and install the module for you. If not, get the tarball, and run these
1811 Construction can be done OO-style (which is the recommended way), or using
1812 Perl's tie() function. Both are examined here.
1814 =head2 OO CONSTRUCTION
1816 The recommended way to construct a DBM::Deep object is to use the new()
1817 method, which gets you a blessed, tied hash or array reference.
1819 my $db = DBM::Deep->new( "foo.db" );
1821 This opens a new database handle, mapped to the file "foo.db". If this
1822 file does not exist, it will automatically be created. DB files are
1823 opened in "r+" (read/write) mode, and the type of object returned is a
1824 hash, unless otherwise specified (see L<OPTIONS> below).
1828 You can pass a number of options to the constructor to specify things like
1829 locking, autoflush, etc. This is done by passing an inline hash:
1831 my $db = DBM::Deep->new(
1837 Notice that the filename is now specified I<inside> the hash with
1838 the "file" parameter, as opposed to being the sole argument to the
1839 constructor. This is required if any options are specified.
1840 See L<OPTIONS> below for the complete list.
1844 You can also start with an array instead of a hash. For this, you must
1845 specify the C<type> parameter:
1847 my $db = DBM::Deep->new(
1849 type => DBM::Deep->TYPE_ARRAY
1852 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1853 a new DB file. If you create a DBM::Deep object with an existing file, the
1854 C<type> will be loaded from the file header, and ignored if it is passed
1857 =head2 TIE CONSTRUCTION
1859 Alternatively, you can create a DBM::Deep handle by using Perl's built-in
1860 tie() function. This is not ideal, because you get only a basic, tied hash
1861 (or array) which is not blessed, so you can't call any functions on it.
1864 tie %hash, "DBM::Deep", "foo.db";
1867 tie @array, "DBM::Deep", "bar.db";
1869 As with the OO constructor, you can replace the DB filename parameter with
1870 a hash containing one or more options (see L<OPTIONS> just below for the
1873 tie %hash, "DBM::Deep", {
1881 There are a number of options that can be passed in when constructing your
1882 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1888 Filename of the DB file to link the handle to. You can pass a full absolute
1889 filesystem path, partial path, or a plain filename if the file is in the
1890 current working directory. This is a required parameter.
1894 File open mode (read-only, read-write, etc.) string passed to Perl's FileHandle
1895 module. This is an optional parameter, and defaults to "r+" (read/write).
1896 B<Note:> If the default (r+) mode is selected, the file will also be auto-
1897 created if it doesn't exist.
1901 This parameter specifies what type of object to create, a hash or array. Use
1902 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1903 This only takes effect when beginning a new file. This is an optional
1904 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1908 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1909 function to lock the database in exclusive mode for writes, and shared mode for
1910 reads. Pass any true value to enable. This affects the base DB handle I<and
1911 any child hashes or arrays> that use the same DB file. This is an optional
1912 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1916 Specifies whether autoflush is to be enabled on the underlying FileHandle.
1917 This obviously slows down write operations, but is required if you may have
1918 multiple processes accessing the same DB file (also consider enable I<locking>
1919 or at least I<volatile>). Pass any true value to enable. This is an optional
1920 parameter, and defaults to 0 (disabled).
1924 If I<volatile> mode is enabled, DBM::Deep will stat() the DB file before each
1925 STORE() operation. This is required if an outside force may change the size of
1926 the file between transactions. Locking also implicitly enables volatile. This
1927 is useful if you want to use a different locking system or write your own. Pass
1928 any true value to enable. This is an optional parameter, and defaults to 0
1933 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1934 restore them when fetched. This is an B<experimental> feature, and does have
1935 side-effects. Basically, when hashes are re-blessed into their original
1936 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1937 able to call any DBM::Deep methods on them. You have been warned.
1938 This is an optional parameter, and defaults to 0 (disabled).
1942 See L<FILTERS> below.
1946 Setting I<debug> mode will make all errors non-fatal, dump them out to
1947 STDERR, and continue on. This is for debugging purposes only, and probably
1948 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1952 Instead of passing a file path, you can instead pass a handle to an pre-opened
1953 filehandle. Note: Beware of using the magick *DATA handle, as this actually
1954 contains your entire Perl script, as well as the data following the __DATA__
1955 marker. This will not work, because DBM::Deep uses absolute seek()s into the
1956 file. Instead, consider reading *DATA into an IO::Scalar handle, then passing
1961 =head1 TIE INTERFACE
1963 With DBM::Deep you can access your databases using Perl's standard hash/array
1964 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can treat
1965 them as such. DBM::Deep will intercept all reads/writes and direct them to the right
1966 place -- the DB file. This has nothing to do with the L<TIE CONSTRUCTION>
1967 section above. This simply tells you how to use DBM::Deep using regular hashes
1968 and arrays, rather than calling functions like C<get()> and C<put()> (although those
1969 work too). It is entirely up to you how to want to access your databases.
1973 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1974 or even nested hashes (or arrays) using standard Perl syntax:
1976 my $db = DBM::Deep->new( "foo.db" );
1978 $db->{mykey} = "myvalue";
1980 $db->{myhash}->{subkey} = "subvalue";
1982 print $db->{myhash}->{subkey} . "\n";
1984 You can even step through hash keys using the normal Perl C<keys()> function:
1986 foreach my $key (keys %$db) {
1987 print "$key: " . $db->{$key} . "\n";
1990 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1991 pushes them onto an array, all before the loop even begins. If you have an
1992 extra large hash, this may exhaust Perl's memory. Instead, consider using
1993 Perl's C<each()> function, which pulls keys/values one at a time, using very
1996 while (my ($key, $value) = each %$db) {
1997 print "$key: $value\n";
2000 Please note that when using C<each()>, you should always pass a direct
2001 hash reference, not a lookup. Meaning, you should B<never> do this:
2004 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
2006 This causes an infinite loop, because for each iteration, Perl is calling
2007 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
2008 it effectively keeps returning the first key over and over again. Instead,
2009 assign a temporary variable to C<$db->{foo}>, then pass that to each().
2013 As with hashes, you can treat any DBM::Deep object like a normal Perl array
2014 reference. This includes inserting, removing and manipulating elements,
2015 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
2016 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
2017 or simply be a nested array reference inside a hash. Example:
2019 my $db = DBM::Deep->new(
2020 file => "foo-array.db",
2021 type => DBM::Deep->TYPE_ARRAY
2025 push @$db, "bar", "baz";
2026 unshift @$db, "bah";
2028 my $last_elem = pop @$db; # baz
2029 my $first_elem = shift @$db; # bah
2030 my $second_elem = $db->[1]; # bar
2032 my $num_elements = scalar @$db;
2036 In addition to the I<tie()> interface, you can also use a standard OO interface
2037 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
2038 array) has its own methods, but both types share the following common methods:
2039 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
2045 Stores a new hash key/value pair, or sets an array element value. Takes two
2046 arguments, the hash key or array index, and the new value. The value can be
2047 a scalar, hash ref or array ref. Returns true on success, false on failure.
2049 $db->put("foo", "bar"); # for hashes
2050 $db->put(1, "bar"); # for arrays
2054 Fetches the value of a hash key or array element. Takes one argument: the hash
2055 key or array index. Returns a scalar, hash ref or array ref, depending on the
2058 my $value = $db->get("foo"); # for hashes
2059 my $value = $db->get(1); # for arrays
2063 Checks if a hash key or array index exists. Takes one argument: the hash key
2064 or array index. Returns true if it exists, false if not.
2066 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
2067 if ($db->exists(1)) { print "yay!\n"; } # for arrays
2071 Deletes one hash key/value pair or array element. Takes one argument: the hash
2072 key or array index. Returns true on success, false if not found. For arrays,
2073 the remaining elements located after the deleted element are NOT moved over.
2074 The deleted element is essentially just undefined, which is exactly how Perl's
2075 internal arrays work. Please note that the space occupied by the deleted
2076 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
2077 below for details and workarounds.
2079 $db->delete("foo"); # for hashes
2080 $db->delete(1); # for arrays
2084 Deletes B<all> hash keys or array elements. Takes no arguments. No return
2085 value. Please note that the space occupied by the deleted keys/values or
2086 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
2087 details and workarounds.
2089 $db->clear(); # hashes or arrays
2095 For hashes, DBM::Deep supports all the common methods described above, and the
2096 following additional methods: C<first_key()> and C<next_key()>.
2102 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
2103 fetched in an undefined order (which appears random). Takes no arguments,
2104 returns the key as a scalar value.
2106 my $key = $db->first_key();
2110 Returns the "next" key in the hash, given the previous one as the sole argument.
2111 Returns undef if there are no more keys to be fetched.
2113 $key = $db->next_key($key);
2117 Here are some examples of using hashes:
2119 my $db = DBM::Deep->new( "foo.db" );
2121 $db->put("foo", "bar");
2122 print "foo: " . $db->get("foo") . "\n";
2124 $db->put("baz", {}); # new child hash ref
2125 $db->get("baz")->put("buz", "biz");
2126 print "buz: " . $db->get("baz")->get("buz") . "\n";
2128 my $key = $db->first_key();
2130 print "$key: " . $db->get($key) . "\n";
2131 $key = $db->next_key($key);
2134 if ($db->exists("foo")) { $db->delete("foo"); }
2138 For arrays, DBM::Deep supports all the common methods described above, and the
2139 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
2140 C<unshift()> and C<splice()>.
2146 Returns the number of elements in the array. Takes no arguments.
2148 my $len = $db->length();
2152 Adds one or more elements onto the end of the array. Accepts scalars, hash
2153 refs or array refs. No return value.
2155 $db->push("foo", "bar", {});
2159 Fetches the last element in the array, and deletes it. Takes no arguments.
2160 Returns undef if array is empty. Returns the element value.
2162 my $elem = $db->pop();
2166 Fetches the first element in the array, deletes it, then shifts all the
2167 remaining elements over to take up the space. Returns the element value. This
2168 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
2171 my $elem = $db->shift();
2175 Inserts one or more elements onto the beginning of the array, shifting all
2176 existing elements over to make room. Accepts scalars, hash refs or array refs.
2177 No return value. This method is not recommended with large arrays -- see
2178 <LARGE ARRAYS> below for details.
2180 $db->unshift("foo", "bar", {});
2184 Performs exactly like Perl's built-in function of the same name. See L<perldoc
2185 -f splice> for usage -- it is too complicated to document here. This method is
2186 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
2190 Here are some examples of using arrays:
2192 my $db = DBM::Deep->new(
2194 type => DBM::Deep->TYPE_ARRAY
2197 $db->push("bar", "baz");
2198 $db->unshift("foo");
2201 my $len = $db->length();
2202 print "length: $len\n"; # 4
2204 for (my $k=0; $k<$len; $k++) {
2205 print "$k: " . $db->get($k) . "\n";
2208 $db->splice(1, 2, "biz", "baf");
2210 while (my $elem = shift @$db) {
2211 print "shifted: $elem\n";
2216 Enable automatic file locking by passing a true value to the C<locking>
2217 parameter when constructing your DBM::Deep object (see L<SETUP> above).
2219 my $db = DBM::Deep->new(
2224 This causes DBM::Deep to C<flock()> the underlying FileHandle object with exclusive
2225 mode for writes, and shared mode for reads. This is required if you have
2226 multiple processes accessing the same database file, to avoid file corruption.
2227 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
2228 NFS> below for more.
2230 =head2 EXPLICIT LOCKING
2232 You can explicitly lock a database, so it remains locked for multiple
2233 transactions. This is done by calling the C<lock()> method, and passing an
2234 optional lock mode argument (defaults to exclusive mode). This is particularly
2235 useful for things like counters, where the current value needs to be fetched,
2236 then incremented, then stored again.
2239 my $counter = $db->get("counter");
2241 $db->put("counter", $counter);
2250 You can pass C<lock()> an optional argument, which specifies which mode to use
2251 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2252 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2253 same as the constants defined in Perl's C<Fcntl> module.
2255 $db->lock( DBM::Deep->LOCK_SH );
2259 If you want to implement your own file locking scheme, be sure to create your
2260 DBM::Deep objects setting the C<volatile> option to true. This hints to DBM::Deep
2261 that the DB file may change between transactions. See L<LOW-LEVEL ACCESS>
2264 =head1 IMPORTING/EXPORTING
2266 You can import existing complex structures by calling the C<import()> method,
2267 and export an entire database into an in-memory structure using the C<export()>
2268 method. Both are examined here.
2272 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2273 walking the structure and adding keys/elements to the database as you go,
2274 simply pass a reference to the C<import()> method. This recursively adds
2275 everything to an existing DBM::Deep object for you. Here is an example:
2280 array1 => [ "elem0", "elem1", "elem2" ],
2282 subkey1 => "subvalue1",
2283 subkey2 => "subvalue2"
2287 my $db = DBM::Deep->new( "foo.db" );
2288 $db->import( $struct );
2290 print $db->{key1} . "\n"; # prints "value1"
2292 This recursively imports the entire C<$struct> object into C<$db>, including
2293 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2294 keys are merged with the existing ones, replacing if they already exist.
2295 The C<import()> method can be called on any database level (not just the base
2296 level), and works with both hash and array DB types.
2300 B<Note:> Make sure your existing structure has no circular references in it.
2301 These will cause an infinite loop when importing.
2305 Calling the C<export()> method on an existing DBM::Deep object will return
2306 a reference to a new in-memory copy of the database. The export is done
2307 recursively, so all nested hashes/arrays are all exported to standard Perl
2308 objects. Here is an example:
2310 my $db = DBM::Deep->new( "foo.db" );
2312 $db->{key1} = "value1";
2313 $db->{key2} = "value2";
2315 $db->{hash1}->{subkey1} = "subvalue1";
2316 $db->{hash1}->{subkey2} = "subvalue2";
2318 my $struct = $db->export();
2320 print $struct->{key1} . "\n"; # prints "value1"
2322 This makes a complete copy of the database in memory, and returns a reference
2323 to it. The C<export()> method can be called on any database level (not just
2324 the base level), and works with both hash and array DB types. Be careful of
2325 large databases -- you can store a lot more data in a DBM::Deep object than an
2326 in-memory Perl structure.
2330 B<Note:> Make sure your database has no circular references in it.
2331 These will cause an infinite loop when exporting.
2335 DBM::Deep has a number of hooks where you can specify your own Perl function
2336 to perform filtering on incoming or outgoing data. This is a perfect
2337 way to extend the engine, and implement things like real-time compression or
2338 encryption. Filtering applies to the base DB level, and all child hashes /
2339 arrays. Filter hooks can be specified when your DBM::Deep object is first
2340 constructed, or by calling the C<set_filter()> method at any time. There are
2341 four available filter hooks, described below:
2345 =item * filter_store_key
2347 This filter is called whenever a hash key is stored. It
2348 is passed the incoming key, and expected to return a transformed key.
2350 =item * filter_store_value
2352 This filter is called whenever a hash key or array element is stored. It
2353 is passed the incoming value, and expected to return a transformed value.
2355 =item * filter_fetch_key
2357 This filter is called whenever a hash key is fetched (i.e. via
2358 C<first_key()> or C<next_key()>). It is passed the transformed key,
2359 and expected to return the plain key.
2361 =item * filter_fetch_value
2363 This filter is called whenever a hash key or array element is fetched.
2364 It is passed the transformed value, and expected to return the plain value.
2368 Here are the two ways to setup a filter hook:
2370 my $db = DBM::Deep->new(
2372 filter_store_value => \&my_filter_store,
2373 filter_fetch_value => \&my_filter_fetch
2378 $db->set_filter( "filter_store_value", \&my_filter_store );
2379 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2381 Your filter function will be called only when dealing with SCALAR keys or
2382 values. When nested hashes and arrays are being stored/fetched, filtering
2383 is bypassed. Filters are called as static functions, passed a single SCALAR
2384 argument, and expected to return a single SCALAR value. If you want to
2385 remove a filter, set the function reference to C<undef>:
2387 $db->set_filter( "filter_store_value", undef );
2389 =head2 REAL-TIME ENCRYPTION EXAMPLE
2391 Here is a working example that uses the I<Crypt::Blowfish> module to
2392 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2393 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2394 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2397 use Crypt::Blowfish;
2400 my $cipher = Crypt::CBC->new({
2401 'key' => 'my secret key',
2402 'cipher' => 'Blowfish',
2404 'regenerate_key' => 0,
2405 'padding' => 'space',
2409 my $db = DBM::Deep->new(
2410 file => "foo-encrypt.db",
2411 filter_store_key => \&my_encrypt,
2412 filter_store_value => \&my_encrypt,
2413 filter_fetch_key => \&my_decrypt,
2414 filter_fetch_value => \&my_decrypt,
2417 $db->{key1} = "value1";
2418 $db->{key2} = "value2";
2419 print "key1: " . $db->{key1} . "\n";
2420 print "key2: " . $db->{key2} . "\n";
2426 return $cipher->encrypt( $_[0] );
2429 return $cipher->decrypt( $_[0] );
2432 =head2 REAL-TIME COMPRESSION EXAMPLE
2434 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2435 compression / decompression of keys & values with DBM::Deep Filters.
2436 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2437 more on I<Compress::Zlib>.
2442 my $db = DBM::Deep->new(
2443 file => "foo-compress.db",
2444 filter_store_key => \&my_compress,
2445 filter_store_value => \&my_compress,
2446 filter_fetch_key => \&my_decompress,
2447 filter_fetch_value => \&my_decompress,
2450 $db->{key1} = "value1";
2451 $db->{key2} = "value2";
2452 print "key1: " . $db->{key1} . "\n";
2453 print "key2: " . $db->{key2} . "\n";
2459 return Compress::Zlib::memGzip( $_[0] ) ;
2462 return Compress::Zlib::memGunzip( $_[0] ) ;
2465 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2466 actually numerical index numbers, and are not filtered.
2468 =head1 ERROR HANDLING
2470 Most DBM::Deep methods return a true value for success, and call die() on
2471 failure. You can wrap calls in an eval block to catch the die. Also, the
2472 actual error message is stored in an internal scalar, which can be fetched by
2473 calling the C<error()> method.
2475 my $db = DBM::Deep->new( "foo.db" ); # create hash
2476 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2478 print $db->error(); # prints error message
2480 You can then call C<clear_error()> to clear the current error state.
2484 If you set the C<debug> option to true when creating your DBM::Deep object,
2485 all errors are considered NON-FATAL, and dumped to STDERR. This is only
2486 for debugging purposes.
2488 =head1 LARGEFILE SUPPORT
2490 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2491 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2492 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2493 by calling the static C<set_pack()> method before you do anything else.
2495 DBM::Deep::set_pack(8, 'Q');
2497 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2498 instead of 32-bit longs. After setting these values your DB files have a
2499 theoretical maximum size of 16 XB (exabytes).
2503 B<Note:> Changing these values will B<NOT> work for existing database files.
2504 Only change this for new files, and make sure it stays set consistently
2505 throughout the file's life. If you do set these values, you can no longer
2506 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2507 back to 32-bit mode.
2511 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2512 only a 32-bit Perl. However, I have received user reports that this does
2515 =head1 LOW-LEVEL ACCESS
2517 If you require low-level access to the underlying FileHandle that DBM::Deep uses,
2518 you can call the C<fh()> method, which returns the handle:
2522 This method can be called on the root level of the datbase, or any child
2523 hashes or arrays. All levels share a I<root> structure, which contains things
2524 like the FileHandle, a reference counter, and all your options you specified
2525 when you created the object. You can get access to this root structure by
2526 calling the C<root()> method.
2528 my $root = $db->root();
2530 This is useful for changing options after the object has already been created,
2531 such as enabling/disabling locking, volatile or debug modes. You can also
2532 store your own temporary user data in this structure (be wary of name
2533 collision), which is then accessible from any child hash or array.
2535 =head1 CUSTOM DIGEST ALGORITHM
2537 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2538 keys. However you can override this, and use another algorithm (such as SHA-256)
2539 or even write your own. But please note that DBM::Deep currently expects zero
2540 collisions, so your algorithm has to be I<perfect>, so to speak.
2541 Collision detection may be introduced in a later version.
2545 You can specify a custom digest algorithm by calling the static C<set_digest()>
2546 function, passing a reference to a subroutine, and the length of the algorithm's
2547 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2548 objects. Here is a working example that uses a 256-bit hash from the
2549 I<Digest::SHA256> module. Please see
2550 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2555 my $context = Digest::SHA256::new(256);
2557 DBM::Deep::set_digest( \&my_digest, 32 );
2559 my $db = DBM::Deep->new( "foo-sha.db" );
2561 $db->{key1} = "value1";
2562 $db->{key2} = "value2";
2563 print "key1: " . $db->{key1} . "\n";
2564 print "key2: " . $db->{key2} . "\n";
2570 return substr( $context->hash($_[0]), 0, 32 );
2573 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2574 of bytes you specify in the C<set_digest()> function (in this case 32).
2576 =head1 CIRCULAR REFERENCES
2578 DBM::Deep has B<experimental> support for circular references. Meaning you
2579 can have a nested hash key or array element that points to a parent object.
2580 This relationship is stored in the DB file, and is preserved between sessions.
2583 my $db = DBM::Deep->new( "foo.db" );
2586 $db->{circle} = $db; # ref to self
2588 print $db->{foo} . "\n"; # prints "foo"
2589 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2591 One catch is, passing the object to a function that recursively walks the
2592 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2593 C<export()> methods) will result in an infinite loop. The other catch is,
2594 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2595 or C<next_key()> methods), you will get the I<target object's key>, not the
2596 ref's key. This gets even more interesting with the above example, where
2597 the I<circle> key points to the base DB object, which technically doesn't
2598 have a key. So I made DBM::Deep return "[base]" as the key name in that
2601 =head1 CAVEATS / ISSUES / BUGS
2603 This section describes all the known issues with DBM::Deep. It you have found
2604 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2606 =head2 UNUSED SPACE RECOVERY
2608 One major caveat with DBM::Deep is that space occupied by existing keys and
2609 values is not recovered when they are deleted. Meaning if you keep deleting
2610 and adding new keys, your file will continuously grow. I am working on this,
2611 but in the meantime you can call the built-in C<optimize()> method from time to
2612 time (perhaps in a crontab or something) to recover all your unused space.
2614 $db->optimize(); # returns true on success
2616 This rebuilds the ENTIRE database into a new file, then moves it on top of
2617 the original. The new file will have no unused space, thus it will take up as
2618 little disk space as possible. Please note that this operation can take
2619 a long time for large files, and you need enough disk space to temporarily hold
2620 2 copies of your DB file. The temporary file is created in the same directory
2621 as the original, named with a ".tmp" extension, and is deleted when the
2622 operation completes. Oh, and if locking is enabled, the DB is automatically
2623 locked for the entire duration of the copy.
2627 B<WARNING:> Only call optimize() on the top-level node of the database, and
2628 make sure there are no child references lying around. DBM::Deep keeps a reference
2629 counter, and if it is greater than 1, optimize() will abort and return undef.
2631 =head2 AUTOVIVIFICATION
2633 Unfortunately, autovivification doesn't work with tied hashes. This appears to
2634 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
2635 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
2636 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
2637 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
2640 $db->{foo}->{bar} = "hello";
2642 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
2643 being an empty hash. Try this instead, which works fine:
2645 $db->{foo} = { bar => "hello" };
2647 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
2648 the execution path, and Perl indeed passes an empty hash to the STORE() method.
2649 Probably a bug in Perl.
2651 =head2 FILE CORRUPTION
2653 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2654 for a 32-bit signature when opened, but other corruption in files can cause
2655 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2656 stuck in an infinite loop depending on the level of corruption. File write
2657 operations are not checked for failure (for speed), so if you happen to run
2658 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2659 be addressed in a later version of DBM::Deep.
2663 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2664 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2665 about setting up your NFS server with a locking daemon, then using lockf() to
2666 lock your files, but your milage may vary there as well. From what I
2667 understand, there is no real way to do it. However, if you need access to the
2668 underlying FileHandle in DBM::Deep for using some other kind of locking scheme like
2669 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2671 =head2 COPYING OBJECTS
2673 Beware of copying tied objects in Perl. Very strange things can happen.
2674 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2675 returns a new, blessed, tied hash or array to the same level in the DB.
2677 my $copy = $db->clone();
2681 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2682 These functions cause every element in the array to move, which can be murder
2683 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2684 a different location. This may be addressed in a later version.
2688 This section discusses DBM::Deep's speed and memory usage.
2692 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2693 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2694 multi-level hash/array support, and cross-platform FTPable files. Even so,
2695 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2696 with huge databases. Here is some test data:
2698 Adding 1,000,000 keys to new DB file...
2700 At 100 keys, avg. speed is 2,703 keys/sec
2701 At 200 keys, avg. speed is 2,642 keys/sec
2702 At 300 keys, avg. speed is 2,598 keys/sec
2703 At 400 keys, avg. speed is 2,578 keys/sec
2704 At 500 keys, avg. speed is 2,722 keys/sec
2705 At 600 keys, avg. speed is 2,628 keys/sec
2706 At 700 keys, avg. speed is 2,700 keys/sec
2707 At 800 keys, avg. speed is 2,607 keys/sec
2708 At 900 keys, avg. speed is 2,190 keys/sec
2709 At 1,000 keys, avg. speed is 2,570 keys/sec
2710 At 2,000 keys, avg. speed is 2,417 keys/sec
2711 At 3,000 keys, avg. speed is 1,982 keys/sec
2712 At 4,000 keys, avg. speed is 1,568 keys/sec
2713 At 5,000 keys, avg. speed is 1,533 keys/sec
2714 At 6,000 keys, avg. speed is 1,787 keys/sec
2715 At 7,000 keys, avg. speed is 1,977 keys/sec
2716 At 8,000 keys, avg. speed is 2,028 keys/sec
2717 At 9,000 keys, avg. speed is 2,077 keys/sec
2718 At 10,000 keys, avg. speed is 2,031 keys/sec
2719 At 20,000 keys, avg. speed is 1,970 keys/sec
2720 At 30,000 keys, avg. speed is 2,050 keys/sec
2721 At 40,000 keys, avg. speed is 2,073 keys/sec
2722 At 50,000 keys, avg. speed is 1,973 keys/sec
2723 At 60,000 keys, avg. speed is 1,914 keys/sec
2724 At 70,000 keys, avg. speed is 2,091 keys/sec
2725 At 80,000 keys, avg. speed is 2,103 keys/sec
2726 At 90,000 keys, avg. speed is 1,886 keys/sec
2727 At 100,000 keys, avg. speed is 1,970 keys/sec
2728 At 200,000 keys, avg. speed is 2,053 keys/sec
2729 At 300,000 keys, avg. speed is 1,697 keys/sec
2730 At 400,000 keys, avg. speed is 1,838 keys/sec
2731 At 500,000 keys, avg. speed is 1,941 keys/sec
2732 At 600,000 keys, avg. speed is 1,930 keys/sec
2733 At 700,000 keys, avg. speed is 1,735 keys/sec
2734 At 800,000 keys, avg. speed is 1,795 keys/sec
2735 At 900,000 keys, avg. speed is 1,221 keys/sec
2736 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2738 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2739 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2740 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2741 Run time was 12 min 3 sec.
2745 One of the great things about DBM::Deep is that it uses very little memory.
2746 Even with huge databases (1,000,000+ keys) you will not see much increased
2747 memory on your process. DBM::Deep relies solely on the filesystem for storing
2748 and fetching data. Here is output from I</usr/bin/top> before even opening a
2751 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2752 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2754 Basically the process is taking 2,716K of memory. And here is the same
2755 process after storing and fetching 1,000,000 keys:
2757 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2758 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2760 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2761 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2763 =head1 DB FILE FORMAT
2765 In case you were interested in the underlying DB file format, it is documented
2766 here in this section. You don't need to know this to use the module, it's just
2767 included for reference.
2771 DBM::Deep files always start with a 32-bit signature to identify the file type.
2772 This is at offset 0. The signature is "DPDB" in network byte order. This is
2773 checked when the file is opened.
2777 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2778 has a standard header containing the type of data, the length of data, and then
2779 the data itself. The type is a single character (1 byte), the length is a
2780 32-bit unsigned long in network byte order, and the data is, well, the data.
2781 Here is how it unfolds:
2785 Immediately after the 32-bit file signature is the I<Master Index> record.
2786 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2787 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2788 depending on how the DBM::Deep object was constructed.
2792 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2793 number). The first 8-bit char of the MD5 signature is the offset into the
2794 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2795 index element is a file offset of the next tag for the key/element in question,
2796 which is usually a I<Bucket List> tag (see below).
2800 The next tag I<could> be another index, depending on how many keys/elements
2801 exist. See L<RE-INDEXING> below for details.
2805 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2806 file offsets to where the actual data is stored. It starts with a standard
2807 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2808 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2809 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2810 When the list fills up, a I<Re-Index> operation is performed (See
2811 L<RE-INDEXING> below).
2815 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2816 index/value pair (in array mode). It starts with a standard tag header with
2817 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2818 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2819 header. The size reported in the tag header is only for the value, but then,
2820 just after the value is another size (32-bit unsigned long) and then the plain
2821 key itself. Since the value is likely to be fetched more often than the plain
2822 key, I figured it would be I<slightly> faster to store the value first.
2826 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2827 record for the nested structure, where the process begins all over again.
2831 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2832 exhausted. Then, when another key/element comes in, the list is converted to a
2833 new index record. However, this index will look at the next char in the MD5
2834 hash, and arrange new Bucket List pointers accordingly. This process is called
2835 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2836 17 (16 + new one) keys/elements are removed from the old Bucket List and
2837 inserted into the new index. Several new Bucket Lists are created in the
2838 process, as a new MD5 char from the key is being examined (it is unlikely that
2839 the keys will all share the same next char of their MD5s).
2843 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2844 when the Bucket Lists will turn into indexes, but the first round tends to
2845 happen right around 4,000 keys. You will see a I<slight> decrease in
2846 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2847 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2848 right around 900,000 keys. This process can continue nearly indefinitely --
2849 right up until the point the I<MD5> signatures start colliding with each other,
2850 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2851 getting struck by lightning while you are walking to cash in your tickets.
2852 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2853 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2854 this is 340 unodecillion, but don't quote me).
2858 When a new key/element is stored, the key (or index number) is first ran through
2859 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2860 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2861 for the first char of the signature (in this case I<b>). If it does not exist,
2862 a new I<Bucket List> is created for our key (and the next 15 future keys that
2863 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2864 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2865 this point, unless we are replacing an existing I<Bucket>), where the actual
2866 data will be stored.
2870 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2871 (or index number), then walking along the indexes. If there are enough
2872 keys/elements in this DB level, there might be nested indexes, each linked to
2873 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2874 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2875 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2876 plain key are stored.
2880 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2881 methods. In this process the indexes are walked systematically, and each key
2882 fetched in increasing MD5 order (which is why it appears random). Once the
2883 I<Bucket> is found, the value is skipped the plain key returned instead.
2884 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2885 alphabetically sorted. This only happens on an index-level -- as soon as the
2886 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2887 so it's pretty much undefined how the keys will come out -- just like Perl's
2890 =head1 CODE COVERAGE
2892 I use B<Devel::Cover> to test the code coverage of my tests, below is the B<Devel::Cover> report on this
2893 module's test suite.
2895 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2896 File stmt bran cond sub pod time total
2897 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2898 blib/lib/DBM/Deep.pm 94.9 84.5 77.8 100.0 11.1 100.0 89.7
2899 Total 94.9 84.5 77.8 100.0 11.1 100.0 89.7
2900 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2904 Joseph Huckaby, L<jhuckaby@cpan.org>
2906 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2910 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2911 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2915 Copyright (c) 2002-2005 Joseph Huckaby. All Rights Reserved.
2916 This is free software, you may use it and distribute it under the
2917 same terms as Perl itself.