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, $DATA_LENGTH_PACK);
63 # Maximum number of buckets per list before another level of indexing is done.
64 # Increase this value for slightly greater speed, but larger database files.
65 # DO NOT decrease this value below 16, due to risk of recursive reindex overrun.
70 # Better not adjust anything below here, unless you're me :-)
74 # Setup digest function for keys
76 our ($DIGEST_FUNC, $HASH_SIZE);
77 #my $DIGEST_FUNC = \&Digest::MD5::md5;
80 # Precalculate index and bucket sizes based on values above.
83 my ($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->_open(); }
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 # Open a fh to the database, create if nonexistent.
209 # Make sure file signature matches DBM::Deep spec.
211 my $self = $_[0]->_get_self;
213 if (defined($self->_fh)) { $self->_close(); }
216 local $SIG{'__DIE__'};
217 # Theoretically, adding O_BINARY should remove the need for the binmode
218 # Of course, testing it is going to be ... interesting.
219 my $flags = O_RDWR | O_CREAT | O_BINARY;
222 sysopen( $fh, $self->_root->{file}, $flags )
224 $self->_root->{fh} = $fh;
225 }; if ($@ ) { $self->_throw_error( "Received error: $@\n" ); }
226 if (! defined($self->_fh)) {
227 return $self->_throw_error("Cannot sysopen file: " . $self->_root->{file} . ": $!");
232 #XXX Can we remove this by using the right sysopen() flags?
233 # Maybe ... q.v. above
234 binmode $fh; # for win32
236 if ($self->_root->{autoflush}) {
237 my $old = select $fh;
242 seek($fh, 0 + $self->_root->{file_offset}, SEEK_SET);
245 my $bytes_read = read( $fh, $signature, length(SIG_FILE));
248 # File is empty -- write signature and master index
251 seek($fh, 0 + $self->_root->{file_offset}, SEEK_SET);
252 print( $fh SIG_FILE);
253 $self->_create_tag($self->_base_offset, $self->_type, chr(0) x $INDEX_SIZE);
255 my $plain_key = "[base]";
256 print( $fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
258 # Flush the filehandle
259 my $old_fh = select $fh;
260 my $old_af = $|; $| = 1; $| = $old_af;
263 my @stats = stat($fh);
264 $self->_root->{inode} = $stats[1];
265 $self->_root->{end} = $stats[7];
271 # Check signature was valid
273 unless ($signature eq SIG_FILE) {
275 return $self->_throw_error("Signature not found -- file is not a Deep DB");
278 my @stats = stat($fh);
279 $self->_root->{inode} = $stats[1];
280 $self->_root->{end} = $stats[7];
283 # Get our type from master index signature
285 my $tag = $self->_load_tag($self->_base_offset);
287 #XXX We probably also want to store the hash algorithm name and not assume anything
288 #XXX The cool thing would be to allow a different hashing algorithm at every level
291 return $self->_throw_error("Corrupted file, no master index record");
293 if ($self->{type} ne $tag->{signature}) {
294 return $self->_throw_error("File type mismatch");
304 my $self = $_[0]->_get_self;
305 close $self->_root->{fh} if $self->_root->{fh};
306 $self->_root->{fh} = undef;
311 # Given offset, signature and content, create tag and write to disk
313 my ($self, $offset, $sig, $content) = @_;
314 my $size = length($content);
318 seek($fh, $offset + $self->_root->{file_offset}, SEEK_SET);
319 print( $fh $sig . pack($DATA_LENGTH_PACK, $size) . $content );
321 if ($offset == $self->_root->{end}) {
322 $self->_root->{end} += SIG_SIZE + $DATA_LENGTH_SIZE + $size;
328 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
335 # Given offset, load single tag and return signature, size and data
342 seek($fh, $offset + $self->_root->{file_offset}, SEEK_SET);
343 if (eof $fh) { return undef; }
346 read( $fh, $b, SIG_SIZE + $DATA_LENGTH_SIZE );
347 my ($sig, $size) = unpack( "A $DATA_LENGTH_PACK", $b );
350 read( $fh, $buffer, $size);
355 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
362 # Given index tag, lookup single entry in index and return .
365 my ($tag, $index) = @_;
367 my $location = unpack($LONG_PACK, substr($tag->{content}, $index * $LONG_SIZE, $LONG_SIZE) );
368 if (!$location) { return; }
370 return $self->_load_tag( $location );
375 # Adds one key/value pair to bucket list, given offset, MD5 digest of key,
376 # plain (undigested) key and value.
379 my ($tag, $md5, $plain_key, $value) = @_;
380 my $keys = $tag->{content};
384 my $root = $self->_root;
386 my $is_dbm_deep = eval { local $SIG{'__DIE__'}; $value->isa( 'DBM::Deep' ) };
387 my $internal_ref = $is_dbm_deep && ($value->_root eq $root);
392 # Iterate through buckets, seeing if this is a new entry or a replace.
394 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
395 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
398 # Found empty bucket (end of list). Populate and exit loop.
402 $location = $internal_ref
403 ? $value->_base_offset
406 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $root->{file_offset}, SEEK_SET);
407 print( $fh $md5 . pack($LONG_PACK, $location) );
411 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
414 # Found existing bucket with same key. Replace with new value.
419 $location = $value->_base_offset;
420 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $root->{file_offset}, SEEK_SET);
421 print( $fh $md5 . pack($LONG_PACK, $location) );
425 seek($fh, $subloc + SIG_SIZE + $root->{file_offset}, SEEK_SET);
427 read( $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' ) {
437 $actual_length = $INDEX_SIZE;
439 # if autobless is enabled, must also take into consideration
440 # the class name, as it is stored along with key/value.
441 if ( $root->{autobless} ) {
442 my $value_class = Scalar::Util::blessed($value);
443 if ( defined $value_class && !$value->isa('DBM::Deep') ) {
444 $actual_length += length($value_class);
448 else { $actual_length = length($value); }
450 if ($actual_length <= $size) {
454 $location = $root->{end};
455 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE + $root->{file_offset}, SEEK_SET);
456 print( $fh pack($LONG_PACK, $location) );
464 # If this is an internal reference, return now.
465 # No need to write value or plain key
472 # If bucket didn't fit into list, split into a new index level
475 seek($fh, $tag->{ref_loc} + $root->{file_offset}, SEEK_SET);
476 print( $fh pack($LONG_PACK, $root->{end}) );
478 my $index_tag = $self->_create_tag($root->{end}, SIG_INDEX, chr(0) x $INDEX_SIZE);
481 $keys .= $md5 . pack($LONG_PACK, 0);
483 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
484 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
486 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
487 my $num = ord(substr($key, $tag->{ch} + 1, 1));
489 if ($offsets[$num]) {
490 my $offset = $offsets[$num] + SIG_SIZE + $DATA_LENGTH_SIZE;
491 seek($fh, $offset + $root->{file_offset}, SEEK_SET);
493 read( $fh, $subkeys, $BUCKET_LIST_SIZE);
495 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
496 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
498 seek($fh, $offset + ($k * $BUCKET_SIZE) + $root->{file_offset}, SEEK_SET);
499 print( $fh $key . pack($LONG_PACK, $old_subloc || $root->{end}) );
505 $offsets[$num] = $root->{end};
506 seek($fh, $index_tag->{offset} + ($num * $LONG_SIZE) + $root->{file_offset}, SEEK_SET);
507 print( $fh pack($LONG_PACK, $root->{end}) );
509 my $blist_tag = $self->_create_tag($root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
511 seek($fh, $blist_tag->{offset} + $root->{file_offset}, SEEK_SET);
512 print( $fh $key . pack($LONG_PACK, $old_subloc || $root->{end}) );
517 $location ||= $root->{end};
518 } # re-index bucket list
521 # Seek to content area and store signature, value and plaintext key
525 seek($fh, $location + $root->{file_offset}, SEEK_SET);
528 # Write signature based on content type, set content length and write actual value.
530 my $r = Scalar::Util::reftype($value) || '';
532 print( $fh TYPE_HASH );
533 print( $fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
534 $content_length = $INDEX_SIZE;
536 elsif ($r eq 'ARRAY') {
537 print( $fh TYPE_ARRAY );
538 print( $fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
539 $content_length = $INDEX_SIZE;
541 elsif (!defined($value)) {
542 print( $fh SIG_NULL );
543 print( $fh pack($DATA_LENGTH_PACK, 0) );
547 print( $fh SIG_DATA );
548 print( $fh pack($DATA_LENGTH_PACK, length($value)) . $value );
549 $content_length = length($value);
553 # Plain key is stored AFTER value, as keys are typically fetched less often.
555 print( $fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
558 # If value is blessed, preserve class name
560 if ( $root->{autobless} ) {
561 my $value_class = Scalar::Util::blessed($value);
562 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
564 # Blessed ref -- will restore later
567 print( $fh pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
568 $content_length += 1;
569 $content_length += $DATA_LENGTH_SIZE + length($value_class);
573 $content_length += 1;
578 # If this is a new content area, advance EOF counter
580 if ($location == $root->{end}) {
581 $root->{end} += SIG_SIZE;
582 $root->{end} += $DATA_LENGTH_SIZE + $content_length;
583 $root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
587 # If content is a hash or array, create new child DBM::Deep object and
588 # pass each key or element to it.
591 my $branch = DBM::Deep->new(
593 base_offset => $location,
596 foreach my $key (keys %{$value}) {
597 $branch->STORE( $key, $value->{$key} );
600 elsif ($r eq 'ARRAY') {
601 my $branch = DBM::Deep->new(
603 base_offset => $location,
607 foreach my $element (@{$value}) {
608 $branch->STORE( $index, $element );
616 return $self->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
619 sub _get_bucket_value {
621 # Fetch single value given tag and MD5 digested key.
624 my ($tag, $md5) = @_;
625 my $keys = $tag->{content};
630 # Iterate through buckets, looking for a key match
633 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
634 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
635 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
639 # Hit end of list, no match
644 if ( $md5 ne $key ) {
649 # Found match -- seek to offset and read signature
652 seek($fh, $subloc + $self->_root->{file_offset}, SEEK_SET);
653 read( $fh, $signature, SIG_SIZE);
656 # If value is a hash or array, return new DBM::Deep object with correct offset
658 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
659 my $obj = DBM::Deep->new(
661 base_offset => $subloc,
665 if ($self->_root->{autobless}) {
667 # Skip over value and plain key to see if object needs
670 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
673 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
674 if ($size) { seek($fh, $size, SEEK_CUR); }
677 read( $fh, $bless_bit, 1);
678 if (ord($bless_bit)) {
680 # Yes, object needs to be re-blessed
683 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
684 if ($size) { read( $fh, $class_name, $size); }
685 if ($class_name) { $obj = bless( $obj, $class_name ); }
693 # Otherwise return actual value
695 elsif ($signature eq SIG_DATA) {
698 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
699 if ($size) { read( $fh, $value, $size); }
704 # Key exists, but content is null
714 # Delete single key/value pair given tag and MD5 digested key.
717 my ($tag, $md5) = @_;
718 my $keys = $tag->{content};
723 # Iterate through buckets, looking for a key match
726 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
727 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
728 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
732 # Hit end of list, no match
737 if ( $md5 ne $key ) {
742 # Matched key -- delete bucket and return
744 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $self->_root->{file_offset}, SEEK_SET);
745 print( $fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
746 print( $fh chr(0) x $BUCKET_SIZE );
756 # Check existence of single key given tag and MD5 digested key.
759 my ($tag, $md5) = @_;
760 my $keys = $tag->{content};
763 # Iterate through buckets, looking for a key match
766 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
767 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
768 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
772 # Hit end of list, no match
777 if ( $md5 ne $key ) {
782 # Matched key -- return true
790 sub _find_bucket_list {
792 # Locate offset for bucket list, given digested key
798 # Locate offset for bucket list using digest index system
801 my $tag = $self->_load_tag($self->_base_offset);
802 if (!$tag) { return; }
804 while ($tag->{signature} ne SIG_BLIST) {
805 $tag = $self->_index_lookup($tag, ord(substr($md5, $ch, 1)));
806 if (!$tag) { return; }
813 sub _traverse_index {
815 # Scan index and recursively step into deeper levels, looking for next key.
817 my ($self, $offset, $ch, $force_return_next) = @_;
818 $force_return_next = undef unless $force_return_next;
820 my $tag = $self->_load_tag( $offset );
824 if ($tag->{signature} ne SIG_BLIST) {
825 my $content = $tag->{content};
827 if ($self->{return_next}) { $start = 0; }
828 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
830 for (my $index = $start; $index < 256; $index++) {
831 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
833 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
834 if (defined($result)) { return $result; }
838 $self->{return_next} = 1;
841 elsif ($tag->{signature} eq SIG_BLIST) {
842 my $keys = $tag->{content};
843 if ($force_return_next) { $self->{return_next} = 1; }
846 # Iterate through buckets, looking for a key match
848 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
849 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
850 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
854 # End of bucket list -- return to outer loop
856 $self->{return_next} = 1;
859 elsif ($key eq $self->{prev_md5}) {
861 # Located previous key -- return next one found
863 $self->{return_next} = 1;
866 elsif ($self->{return_next}) {
868 # Seek to bucket location and skip over signature
870 seek($fh, $subloc + SIG_SIZE + $self->_root->{file_offset}, SEEK_SET);
873 # Skip over value to get to plain key
876 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
877 if ($size) { seek($fh, $size, SEEK_CUR); }
880 # Read in plain key and return as scalar
883 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
884 if ($size) { read( $fh, $plain_key, $size); }
890 $self->{return_next} = 1;
891 } # tag is a bucket list
898 # Locate next key, given digested previous one
900 my $self = $_[0]->_get_self;
902 $self->{prev_md5} = $_[1] ? $_[1] : undef;
903 $self->{return_next} = 0;
906 # If the previous key was not specifed, start at the top and
907 # return the first one found.
909 if (!$self->{prev_md5}) {
910 $self->{prev_md5} = chr(0) x $HASH_SIZE;
911 $self->{return_next} = 1;
914 return $self->_traverse_index( $self->_base_offset, 0 );
919 # If db locking is set, flock() the db file. If called multiple
920 # times before unlock(), then the same number of unlocks() must
921 # be called before the lock is released.
923 my $self = $_[0]->_get_self;
925 $type = LOCK_EX unless defined $type;
927 if (!defined($self->_fh)) { return; }
929 if ($self->_root->{locking}) {
930 if (!$self->_root->{locked}) {
931 flock($self->_fh, $type);
933 # refresh end counter in case file has changed size
934 my @stats = stat($self->_root->{file});
935 $self->_root->{end} = $stats[7];
937 # double-check file inode, in case another process
938 # has optimize()d our file while we were waiting.
939 if ($stats[1] != $self->_root->{inode}) {
940 $self->_open(); # re-open
941 flock($self->_fh, $type); # re-lock
942 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
945 $self->_root->{locked}++;
955 # If db locking is set, unlock the db file. See note in lock()
956 # regarding calling lock() multiple times.
958 my $self = $_[0]->_get_self;
960 if (!defined($self->_fh)) { return; }
962 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
963 $self->_root->{locked}--;
964 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
973 my $self = shift->_get_self;
974 my ($spot, $value) = @_;
979 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
980 my $type = $value->_type;
981 ${$spot} = $type eq TYPE_HASH ? {} : [];
982 $value->_copy_node( ${$spot} );
985 my $r = Scalar::Util::reftype( $value );
986 my $c = Scalar::Util::blessed( $value );
987 if ( $r eq 'ARRAY' ) {
988 ${$spot} = [ @{$value} ];
991 ${$spot} = { %{$value} };
993 ${$spot} = bless ${$spot}, $c
1002 # Copy single level of keys or elements to new DB handle.
1003 # Recurse for nested structures
1005 my $self = shift->_get_self;
1008 if ($self->_type eq TYPE_HASH) {
1009 my $key = $self->first_key();
1011 my $value = $self->get($key);
1012 $self->_copy_value( \$db_temp->{$key}, $value );
1013 $key = $self->next_key($key);
1017 my $length = $self->length();
1018 for (my $index = 0; $index < $length; $index++) {
1019 my $value = $self->get($index);
1020 $self->_copy_value( \$db_temp->[$index], $value );
1029 # Recursively export into standard Perl hashes and arrays.
1031 my $self = $_[0]->_get_self;
1034 if ($self->_type eq TYPE_HASH) { $temp = {}; }
1035 elsif ($self->_type eq TYPE_ARRAY) { $temp = []; }
1038 $self->_copy_node( $temp );
1046 # Recursively import Perl hash/array structure
1048 #XXX This use of ref() seems to be ok
1049 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
1051 my $self = $_[0]->_get_self;
1054 #XXX This use of ref() seems to be ok
1055 if (!ref($struct)) {
1057 # struct is not a reference, so just import based on our type
1061 if ($self->_type eq TYPE_HASH) { $struct = {@_}; }
1062 elsif ($self->_type eq TYPE_ARRAY) { $struct = [@_]; }
1065 my $r = Scalar::Util::reftype($struct) || '';
1066 if ($r eq "HASH" && $self->_type eq TYPE_HASH) {
1067 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
1069 elsif ($r eq "ARRAY" && $self->_type eq TYPE_ARRAY) {
1070 $self->push( @$struct );
1073 return $self->_throw_error("Cannot import: type mismatch");
1081 # Rebuild entire database into new file, then move
1082 # it back on top of original.
1084 my $self = $_[0]->_get_self;
1086 #XXX Need to create a new test for this
1087 # if ($self->_root->{links} > 1) {
1088 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
1091 my $db_temp = DBM::Deep->new(
1092 file => $self->_root->{file} . '.tmp',
1093 type => $self->_type
1096 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
1100 $self->_copy_node( $db_temp );
1104 # Attempt to copy user, group and permissions over to new file
1106 my @stats = stat($self->_fh);
1107 my $perms = $stats[2] & 07777;
1108 my $uid = $stats[4];
1109 my $gid = $stats[5];
1110 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
1111 chmod( $perms, $self->_root->{file} . '.tmp' );
1113 # q.v. perlport for more information on this variable
1114 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
1116 # Potential race condition when optmizing on Win32 with locking.
1117 # The Windows filesystem requires that the filehandle be closed
1118 # before it is overwritten with rename(). This could be redone
1125 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
1126 unlink $self->_root->{file} . '.tmp';
1128 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
1140 # Make copy of object and return
1142 my $self = $_[0]->_get_self;
1144 return DBM::Deep->new(
1145 type => $self->_type,
1146 base_offset => $self->_base_offset,
1147 root => $self->_root
1152 my %is_legal_filter = map {
1155 store_key store_value
1156 fetch_key fetch_value
1161 # Setup filter function for storing or fetching the key or value
1163 my $self = $_[0]->_get_self;
1164 my $type = lc $_[1];
1165 my $func = $_[2] ? $_[2] : undef;
1167 if ( $is_legal_filter{$type} ) {
1168 $self->_root->{"filter_$type"} = $func;
1182 # Get access to the root structure
1184 my $self = $_[0]->_get_self;
1185 return $self->{root};
1190 # Get access to the raw fh
1192 #XXX It will be useful, though, when we split out HASH and ARRAY
1193 my $self = $_[0]->_get_self;
1194 return $self->_root->{fh};
1199 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
1201 my $self = $_[0]->_get_self;
1202 return $self->{type};
1207 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
1209 my $self = $_[0]->_get_self;
1210 return $self->{base_offset};
1218 die "DBM::Deep: $_[1]\n";
1221 sub _precalc_sizes {
1223 # Precalculate index, bucket and bucket list sizes
1226 #XXX I don't like this ...
1227 set_pack() unless defined $LONG_SIZE;
1229 $INDEX_SIZE = 256 * $LONG_SIZE;
1230 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
1231 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
1236 # Set pack/unpack modes (see file header for more)
1238 my ($long_s, $long_p, $data_s, $data_p) = @_;
1240 $LONG_SIZE = $long_s ? $long_s : 4;
1241 $LONG_PACK = $long_p ? $long_p : 'N';
1243 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
1244 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
1251 # Set key digest function (default is MD5)
1253 my ($digest_func, $hash_size) = @_;
1255 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
1256 $HASH_SIZE = $hash_size ? $hash_size : 16;
1263 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
1268 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
1272 # tie() methods (hashes and arrays)
1277 # Store single hash key/value or array element in database.
1279 my $self = $_[0]->_get_self;
1282 # User may be storing a hash, in which case we do not want it run
1283 # through the filtering system
1284 my $value = ($self->_root->{filter_store_value} && !ref($_[2]))
1285 ? $self->_root->{filter_store_value}->($_[2])
1288 my $md5 = $DIGEST_FUNC->($key);
1291 # Make sure file is open
1293 # if (!defined($self->_fh) && !$self->_open()) {
1297 unless ( _is_writable( $self->_fh ) ) {
1298 $self->_throw_error( 'Cannot write to a readonly filehandle' );
1302 # Request exclusive lock for writing
1304 $self->lock( LOCK_EX );
1306 my $fh = $self->_fh;
1309 # Locate offset for bucket list using digest index system
1311 my $tag = $self->_load_tag($self->_base_offset);
1313 $tag = $self->_create_tag($self->_base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1317 while ($tag->{signature} ne SIG_BLIST) {
1318 my $num = ord(substr($md5, $ch, 1));
1320 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1321 my $new_tag = $self->_index_lookup($tag, $num);
1324 seek($fh, $ref_loc + $self->_root->{file_offset}, SEEK_SET);
1325 print( $fh pack($LONG_PACK, $self->_root->{end}) );
1327 $tag = $self->_create_tag($self->_root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1329 $tag->{ref_loc} = $ref_loc;
1337 $tag->{ref_loc} = $ref_loc;
1344 # Add key/value to bucket list
1346 my $result = $self->_add_bucket( $tag, $md5, $key, $value );
1355 # Fetch single value or element given plain key or array index
1357 my $self = shift->_get_self;
1361 # Make sure file is open
1363 # if (!defined($self->_fh)) { $self->_open(); }
1365 my $md5 = $DIGEST_FUNC->($key);
1368 # Request shared lock for reading
1370 $self->lock( LOCK_SH );
1372 my $tag = $self->_find_bucket_list( $md5 );
1379 # Get value from bucket list
1381 my $result = $self->_get_bucket_value( $tag, $md5 );
1385 #XXX What is ref() checking here?
1386 #YYY Filters only apply on scalar values, so the ref check is making
1387 #YYY sure the fetched bucket is a scalar, not a child hash or array.
1388 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
1389 ? $self->_root->{filter_fetch_value}->($result)
1395 # Delete single key/value pair or element given plain key or array index
1397 my $self = $_[0]->_get_self;
1400 my $md5 = $DIGEST_FUNC->($key);
1403 # Make sure file is open
1405 # if (!defined($self->_fh)) { $self->_open(); }
1408 # Request exclusive lock for writing
1410 $self->lock( LOCK_EX );
1412 my $tag = $self->_find_bucket_list( $md5 );
1421 my $value = $self->_get_bucket_value( $tag, $md5 );
1422 if ($value && !ref($value) && $self->_root->{filter_fetch_value}) {
1423 $value = $self->_root->{filter_fetch_value}->($value);
1426 my $result = $self->_delete_bucket( $tag, $md5 );
1429 # If this object is an array and the key deleted was on the end of the stack,
1430 # decrement the length variable.
1440 # Check if a single key or element exists given plain key or array index
1442 my $self = $_[0]->_get_self;
1445 my $md5 = $DIGEST_FUNC->($key);
1448 # Make sure file is open
1450 # if (!defined($self->_fh)) { $self->_open(); }
1453 # Request shared lock for reading
1455 $self->lock( LOCK_SH );
1457 my $tag = $self->_find_bucket_list( $md5 );
1460 # For some reason, the built-in exists() function returns '' for false
1468 # Check if bucket exists and return 1 or ''
1470 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1479 # Clear all keys from hash, or all elements from array.
1481 my $self = $_[0]->_get_self;
1484 # Make sure file is open
1486 # if (!defined($self->_fh)) { $self->_open(); }
1489 # Request exclusive lock for writing
1491 $self->lock( LOCK_EX );
1493 my $fh = $self->_fh;
1495 seek($fh, $self->_base_offset + $self->_root->{file_offset}, SEEK_SET);
1501 $self->_create_tag($self->_base_offset, $self->_type, chr(0) x $INDEX_SIZE);
1509 # Public method aliases
1511 sub put { (shift)->STORE( @_ ) }
1512 sub store { (shift)->STORE( @_ ) }
1513 sub get { (shift)->FETCH( @_ ) }
1514 sub fetch { (shift)->FETCH( @_ ) }
1515 sub delete { (shift)->DELETE( @_ ) }
1516 sub exists { (shift)->EXISTS( @_ ) }
1517 sub clear { (shift)->CLEAR( @_ ) }
1519 package DBM::Deep::_::Root;
1533 filter_store_key => undef,
1534 filter_store_value => undef,
1535 filter_fetch_key => undef,
1536 filter_fetch_value => undef,
1542 if ( $self->{fh} && !$self->{file_offset} ) {
1543 $self->{file_offset} = tell( $self->{fh} );
1551 return unless $self;
1553 close $self->{fh} if $self->{fh};
1564 DBM::Deep - A pure perl multi-level hash/array DBM
1569 my $db = DBM::Deep->new( "foo.db" );
1571 $db->{key} = 'value'; # tie() style
1574 $db->put('key' => 'value'); # OO style
1575 print $db->get('key');
1577 # true multi-level support
1578 $db->{my_complex} = [
1579 'hello', { perl => 'rules' },
1585 A unique flat-file database module, written in pure perl. True
1586 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1587 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1588 handle millions of keys and unlimited hash levels without significant
1589 slow-down. Written from the ground-up in pure perl -- this is NOT a
1590 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1591 Mac OS X and Windows.
1595 Hopefully you are using Perl's excellent CPAN module, which will download
1596 and install the module for you. If not, get the tarball, and run these
1608 Construction can be done OO-style (which is the recommended way), or using
1609 Perl's tie() function. Both are examined here.
1611 =head2 OO CONSTRUCTION
1613 The recommended way to construct a DBM::Deep object is to use the new()
1614 method, which gets you a blessed, tied hash or array reference.
1616 my $db = DBM::Deep->new( "foo.db" );
1618 This opens a new database handle, mapped to the file "foo.db". If this
1619 file does not exist, it will automatically be created. DB files are
1620 opened in "r+" (read/write) mode, and the type of object returned is a
1621 hash, unless otherwise specified (see L<OPTIONS> below).
1623 You can pass a number of options to the constructor to specify things like
1624 locking, autoflush, etc. This is done by passing an inline hash:
1626 my $db = DBM::Deep->new(
1632 Notice that the filename is now specified I<inside> the hash with
1633 the "file" parameter, as opposed to being the sole argument to the
1634 constructor. This is required if any options are specified.
1635 See L<OPTIONS> below for the complete list.
1639 You can also start with an array instead of a hash. For this, you must
1640 specify the C<type> parameter:
1642 my $db = DBM::Deep->new(
1644 type => DBM::Deep->TYPE_ARRAY
1647 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1648 a new DB file. If you create a DBM::Deep object with an existing file, the
1649 C<type> will be loaded from the file header, and an error will be thrown if
1650 the wrong type is passed in.
1652 =head2 TIE CONSTRUCTION
1654 Alternately, you can create a DBM::Deep handle by using Perl's built-in
1655 tie() function. The object returned from tie() can be used to call methods,
1656 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
1657 file (as expected with most tie'd objects).
1660 my $db = tie %hash, "DBM::Deep", "foo.db";
1663 my $db = tie @array, "DBM::Deep", "bar.db";
1665 As with the OO constructor, you can replace the DB filename parameter with
1666 a hash containing one or more options (see L<OPTIONS> just below for the
1669 tie %hash, "DBM::Deep", {
1677 There are a number of options that can be passed in when constructing your
1678 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1684 Filename of the DB file to link the handle to. You can pass a full absolute
1685 filesystem path, partial path, or a plain filename if the file is in the
1686 current working directory. This is a required parameter (though q.v. fh).
1690 If you want, you can pass in the fh instead of the file. This is most useful for doing
1693 my $db = DBM::Deep->new( { fh => \*DATA } );
1695 You are responsible for making sure that the fh has been opened appropriately for your
1696 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
1697 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
1698 needs to read from the fh.
1702 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
1703 not need to set this. However, it's there if you want it.
1705 If you pass in fh and do not set this, it will be set appropriately.
1709 This parameter specifies what type of object to create, a hash or array. Use
1710 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1711 This only takes effect when beginning a new file. This is an optional
1712 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1716 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1717 function to lock the database in exclusive mode for writes, and shared mode for
1718 reads. Pass any true value to enable. This affects the base DB handle I<and
1719 any child hashes or arrays> that use the same DB file. This is an optional
1720 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1724 Specifies whether autoflush is to be enabled on the underlying filehandle.
1725 This obviously slows down write operations, but is required if you may have
1726 multiple processes accessing the same DB file (also consider enable I<locking>).
1727 Pass any true value to enable. This is an optional parameter, and defaults to 0
1732 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1733 restore them when fetched. This is an B<experimental> feature, and does have
1734 side-effects. Basically, when hashes are re-blessed into their original
1735 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1736 able to call any DBM::Deep methods on them. You have been warned.
1737 This is an optional parameter, and defaults to 0 (disabled).
1741 See L<FILTERS> below.
1745 Setting I<debug> mode will make all errors non-fatal, dump them out to
1746 STDERR, and continue on. This is for debugging purposes only, and probably
1747 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1749 B<NOTE>: This parameter is considered deprecated and should not be used anymore.
1753 =head1 TIE INTERFACE
1755 With DBM::Deep you can access your databases using Perl's standard hash/array
1756 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
1757 treat them as such. DBM::Deep will intercept all reads/writes and direct them
1758 to the right place -- the DB file. This has nothing to do with the
1759 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
1760 using regular hashes and arrays, rather than calling functions like C<get()>
1761 and C<put()> (although those work too). It is entirely up to you how to want
1762 to access your databases.
1766 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1767 or even nested hashes (or arrays) using standard Perl syntax:
1769 my $db = DBM::Deep->new( "foo.db" );
1771 $db->{mykey} = "myvalue";
1773 $db->{myhash}->{subkey} = "subvalue";
1775 print $db->{myhash}->{subkey} . "\n";
1777 You can even step through hash keys using the normal Perl C<keys()> function:
1779 foreach my $key (keys %$db) {
1780 print "$key: " . $db->{$key} . "\n";
1783 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1784 pushes them onto an array, all before the loop even begins. If you have an
1785 extra large hash, this may exhaust Perl's memory. Instead, consider using
1786 Perl's C<each()> function, which pulls keys/values one at a time, using very
1789 while (my ($key, $value) = each %$db) {
1790 print "$key: $value\n";
1793 Please note that when using C<each()>, you should always pass a direct
1794 hash reference, not a lookup. Meaning, you should B<never> do this:
1797 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1799 This causes an infinite loop, because for each iteration, Perl is calling
1800 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1801 it effectively keeps returning the first key over and over again. Instead,
1802 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1806 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1807 reference. This includes inserting, removing and manipulating elements,
1808 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1809 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1810 or simply be a nested array reference inside a hash. Example:
1812 my $db = DBM::Deep->new(
1813 file => "foo-array.db",
1814 type => DBM::Deep->TYPE_ARRAY
1818 push @$db, "bar", "baz";
1819 unshift @$db, "bah";
1821 my $last_elem = pop @$db; # baz
1822 my $first_elem = shift @$db; # bah
1823 my $second_elem = $db->[1]; # bar
1825 my $num_elements = scalar @$db;
1829 In addition to the I<tie()> interface, you can also use a standard OO interface
1830 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1831 array) has its own methods, but both types share the following common methods:
1832 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1836 =item * new() / clone()
1838 These are the constructor and copy-functions.
1840 =item * put() / store()
1842 Stores a new hash key/value pair, or sets an array element value. Takes two
1843 arguments, the hash key or array index, and the new value. The value can be
1844 a scalar, hash ref or array ref. Returns true on success, false on failure.
1846 $db->put("foo", "bar"); # for hashes
1847 $db->put(1, "bar"); # for arrays
1849 =item * get() / fetch()
1851 Fetches the value of a hash key or array element. Takes one argument: the hash
1852 key or array index. Returns a scalar, hash ref or array ref, depending on the
1855 my $value = $db->get("foo"); # for hashes
1856 my $value = $db->get(1); # for arrays
1860 Checks if a hash key or array index exists. Takes one argument: the hash key
1861 or array index. Returns true if it exists, false if not.
1863 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1864 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1868 Deletes one hash key/value pair or array element. Takes one argument: the hash
1869 key or array index. Returns true on success, false if not found. For arrays,
1870 the remaining elements located after the deleted element are NOT moved over.
1871 The deleted element is essentially just undefined, which is exactly how Perl's
1872 internal arrays work. Please note that the space occupied by the deleted
1873 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1874 below for details and workarounds.
1876 $db->delete("foo"); # for hashes
1877 $db->delete(1); # for arrays
1881 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1882 value. Please note that the space occupied by the deleted keys/values or
1883 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1884 details and workarounds.
1886 $db->clear(); # hashes or arrays
1888 =item * lock() / unlock()
1894 Recover lost disk space.
1896 =item * import() / export()
1898 Data going in and out.
1900 =item * set_digest() / set_pack() / set_filter()
1902 q.v. adjusting the interal parameters.
1908 For hashes, DBM::Deep supports all the common methods described above, and the
1909 following additional methods: C<first_key()> and C<next_key()>.
1915 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1916 fetched in an undefined order (which appears random). Takes no arguments,
1917 returns the key as a scalar value.
1919 my $key = $db->first_key();
1923 Returns the "next" key in the hash, given the previous one as the sole argument.
1924 Returns undef if there are no more keys to be fetched.
1926 $key = $db->next_key($key);
1930 Here are some examples of using hashes:
1932 my $db = DBM::Deep->new( "foo.db" );
1934 $db->put("foo", "bar");
1935 print "foo: " . $db->get("foo") . "\n";
1937 $db->put("baz", {}); # new child hash ref
1938 $db->get("baz")->put("buz", "biz");
1939 print "buz: " . $db->get("baz")->get("buz") . "\n";
1941 my $key = $db->first_key();
1943 print "$key: " . $db->get($key) . "\n";
1944 $key = $db->next_key($key);
1947 if ($db->exists("foo")) { $db->delete("foo"); }
1951 For arrays, DBM::Deep supports all the common methods described above, and the
1952 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1953 C<unshift()> and C<splice()>.
1959 Returns the number of elements in the array. Takes no arguments.
1961 my $len = $db->length();
1965 Adds one or more elements onto the end of the array. Accepts scalars, hash
1966 refs or array refs. No return value.
1968 $db->push("foo", "bar", {});
1972 Fetches the last element in the array, and deletes it. Takes no arguments.
1973 Returns undef if array is empty. Returns the element value.
1975 my $elem = $db->pop();
1979 Fetches the first element in the array, deletes it, then shifts all the
1980 remaining elements over to take up the space. Returns the element value. This
1981 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1984 my $elem = $db->shift();
1988 Inserts one or more elements onto the beginning of the array, shifting all
1989 existing elements over to make room. Accepts scalars, hash refs or array refs.
1990 No return value. This method is not recommended with large arrays -- see
1991 <LARGE ARRAYS> below for details.
1993 $db->unshift("foo", "bar", {});
1997 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1998 -f splice> for usage -- it is too complicated to document here. This method is
1999 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
2003 Here are some examples of using arrays:
2005 my $db = DBM::Deep->new(
2007 type => DBM::Deep->TYPE_ARRAY
2010 $db->push("bar", "baz");
2011 $db->unshift("foo");
2014 my $len = $db->length();
2015 print "length: $len\n"; # 4
2017 for (my $k=0; $k<$len; $k++) {
2018 print "$k: " . $db->get($k) . "\n";
2021 $db->splice(1, 2, "biz", "baf");
2023 while (my $elem = shift @$db) {
2024 print "shifted: $elem\n";
2029 Enable automatic file locking by passing a true value to the C<locking>
2030 parameter when constructing your DBM::Deep object (see L<SETUP> above).
2032 my $db = DBM::Deep->new(
2037 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
2038 mode for writes, and shared mode for reads. This is required if you have
2039 multiple processes accessing the same database file, to avoid file corruption.
2040 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
2041 NFS> below for more.
2043 =head2 EXPLICIT LOCKING
2045 You can explicitly lock a database, so it remains locked for multiple
2046 transactions. This is done by calling the C<lock()> method, and passing an
2047 optional lock mode argument (defaults to exclusive mode). This is particularly
2048 useful for things like counters, where the current value needs to be fetched,
2049 then incremented, then stored again.
2052 my $counter = $db->get("counter");
2054 $db->put("counter", $counter);
2063 You can pass C<lock()> an optional argument, which specifies which mode to use
2064 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2065 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2066 same as the constants defined in Perl's C<Fcntl> module.
2068 $db->lock( DBM::Deep->LOCK_SH );
2072 =head1 IMPORTING/EXPORTING
2074 You can import existing complex structures by calling the C<import()> method,
2075 and export an entire database into an in-memory structure using the C<export()>
2076 method. Both are examined here.
2080 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2081 walking the structure and adding keys/elements to the database as you go,
2082 simply pass a reference to the C<import()> method. This recursively adds
2083 everything to an existing DBM::Deep object for you. Here is an example:
2088 array1 => [ "elem0", "elem1", "elem2" ],
2090 subkey1 => "subvalue1",
2091 subkey2 => "subvalue2"
2095 my $db = DBM::Deep->new( "foo.db" );
2096 $db->import( $struct );
2098 print $db->{key1} . "\n"; # prints "value1"
2100 This recursively imports the entire C<$struct> object into C<$db>, including
2101 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2102 keys are merged with the existing ones, replacing if they already exist.
2103 The C<import()> method can be called on any database level (not just the base
2104 level), and works with both hash and array DB types.
2106 B<Note:> Make sure your existing structure has no circular references in it.
2107 These will cause an infinite loop when importing.
2111 Calling the C<export()> method on an existing DBM::Deep object will return
2112 a reference to a new in-memory copy of the database. The export is done
2113 recursively, so all nested hashes/arrays are all exported to standard Perl
2114 objects. Here is an example:
2116 my $db = DBM::Deep->new( "foo.db" );
2118 $db->{key1} = "value1";
2119 $db->{key2} = "value2";
2121 $db->{hash1}->{subkey1} = "subvalue1";
2122 $db->{hash1}->{subkey2} = "subvalue2";
2124 my $struct = $db->export();
2126 print $struct->{key1} . "\n"; # prints "value1"
2128 This makes a complete copy of the database in memory, and returns a reference
2129 to it. The C<export()> method can be called on any database level (not just
2130 the base level), and works with both hash and array DB types. Be careful of
2131 large databases -- you can store a lot more data in a DBM::Deep object than an
2132 in-memory Perl structure.
2134 B<Note:> Make sure your database has no circular references in it.
2135 These will cause an infinite loop when exporting.
2139 DBM::Deep has a number of hooks where you can specify your own Perl function
2140 to perform filtering on incoming or outgoing data. This is a perfect
2141 way to extend the engine, and implement things like real-time compression or
2142 encryption. Filtering applies to the base DB level, and all child hashes /
2143 arrays. Filter hooks can be specified when your DBM::Deep object is first
2144 constructed, or by calling the C<set_filter()> method at any time. There are
2145 four available filter hooks, described below:
2149 =item * filter_store_key
2151 This filter is called whenever a hash key is stored. It
2152 is passed the incoming key, and expected to return a transformed key.
2154 =item * filter_store_value
2156 This filter is called whenever a hash key or array element is stored. It
2157 is passed the incoming value, and expected to return a transformed value.
2159 =item * filter_fetch_key
2161 This filter is called whenever a hash key is fetched (i.e. via
2162 C<first_key()> or C<next_key()>). It is passed the transformed key,
2163 and expected to return the plain key.
2165 =item * filter_fetch_value
2167 This filter is called whenever a hash key or array element is fetched.
2168 It is passed the transformed value, and expected to return the plain value.
2172 Here are the two ways to setup a filter hook:
2174 my $db = DBM::Deep->new(
2176 filter_store_value => \&my_filter_store,
2177 filter_fetch_value => \&my_filter_fetch
2182 $db->set_filter( "filter_store_value", \&my_filter_store );
2183 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2185 Your filter function will be called only when dealing with SCALAR keys or
2186 values. When nested hashes and arrays are being stored/fetched, filtering
2187 is bypassed. Filters are called as static functions, passed a single SCALAR
2188 argument, and expected to return a single SCALAR value. If you want to
2189 remove a filter, set the function reference to C<undef>:
2191 $db->set_filter( "filter_store_value", undef );
2193 =head2 REAL-TIME ENCRYPTION EXAMPLE
2195 Here is a working example that uses the I<Crypt::Blowfish> module to
2196 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2197 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2198 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2201 use Crypt::Blowfish;
2204 my $cipher = Crypt::CBC->new({
2205 'key' => 'my secret key',
2206 'cipher' => 'Blowfish',
2208 'regenerate_key' => 0,
2209 'padding' => 'space',
2213 my $db = DBM::Deep->new(
2214 file => "foo-encrypt.db",
2215 filter_store_key => \&my_encrypt,
2216 filter_store_value => \&my_encrypt,
2217 filter_fetch_key => \&my_decrypt,
2218 filter_fetch_value => \&my_decrypt,
2221 $db->{key1} = "value1";
2222 $db->{key2} = "value2";
2223 print "key1: " . $db->{key1} . "\n";
2224 print "key2: " . $db->{key2} . "\n";
2230 return $cipher->encrypt( $_[0] );
2233 return $cipher->decrypt( $_[0] );
2236 =head2 REAL-TIME COMPRESSION EXAMPLE
2238 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2239 compression / decompression of keys & values with DBM::Deep Filters.
2240 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2241 more on I<Compress::Zlib>.
2246 my $db = DBM::Deep->new(
2247 file => "foo-compress.db",
2248 filter_store_key => \&my_compress,
2249 filter_store_value => \&my_compress,
2250 filter_fetch_key => \&my_decompress,
2251 filter_fetch_value => \&my_decompress,
2254 $db->{key1} = "value1";
2255 $db->{key2} = "value2";
2256 print "key1: " . $db->{key1} . "\n";
2257 print "key2: " . $db->{key2} . "\n";
2263 return Compress::Zlib::memGzip( $_[0] ) ;
2266 return Compress::Zlib::memGunzip( $_[0] ) ;
2269 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2270 actually numerical index numbers, and are not filtered.
2272 =head1 ERROR HANDLING
2274 Most DBM::Deep methods return a true value for success, and call die() on
2275 failure. You can wrap calls in an eval block to catch the die.
2277 my $db = DBM::Deep->new( "foo.db" ); # create hash
2278 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2280 print $@; # prints error message
2282 =head1 LARGEFILE SUPPORT
2284 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2285 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2286 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2287 by calling the static C<set_pack()> method before you do anything else.
2289 DBM::Deep::set_pack(8, 'Q');
2291 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2292 instead of 32-bit longs. After setting these values your DB files have a
2293 theoretical maximum size of 16 XB (exabytes).
2295 B<Note:> Changing these values will B<NOT> work for existing database files.
2296 Only change this for new files, and make sure it stays set consistently
2297 throughout the file's life. If you do set these values, you can no longer
2298 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2299 back to 32-bit mode.
2301 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2302 only a 32-bit Perl. However, I have received user reports that this does
2305 =head1 LOW-LEVEL ACCESS
2307 If you require low-level access to the underlying filehandle that DBM::Deep uses,
2308 you can call the C<_fh()> method, which returns the handle:
2310 my $fh = $db->_fh();
2312 This method can be called on the root level of the datbase, or any child
2313 hashes or arrays. All levels share a I<root> structure, which contains things
2314 like the filehandle, a reference counter, and all the options specified
2315 when you created the object. You can get access to this root structure by
2316 calling the C<root()> method.
2318 my $root = $db->_root();
2320 This is useful for changing options after the object has already been created,
2321 such as enabling/disabling locking, or debug modes. You can also
2322 store your own temporary user data in this structure (be wary of name
2323 collision), which is then accessible from any child hash or array.
2325 =head1 CUSTOM DIGEST ALGORITHM
2327 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2328 keys. However you can override this, and use another algorithm (such as SHA-256)
2329 or even write your own. But please note that DBM::Deep currently expects zero
2330 collisions, so your algorithm has to be I<perfect>, so to speak.
2331 Collision detection may be introduced in a later version.
2335 You can specify a custom digest algorithm by calling the static C<set_digest()>
2336 function, passing a reference to a subroutine, and the length of the algorithm's
2337 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2338 objects. Here is a working example that uses a 256-bit hash from the
2339 I<Digest::SHA256> module. Please see
2340 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2345 my $context = Digest::SHA256::new(256);
2347 DBM::Deep::set_digest( \&my_digest, 32 );
2349 my $db = DBM::Deep->new( "foo-sha.db" );
2351 $db->{key1} = "value1";
2352 $db->{key2} = "value2";
2353 print "key1: " . $db->{key1} . "\n";
2354 print "key2: " . $db->{key2} . "\n";
2360 return substr( $context->hash($_[0]), 0, 32 );
2363 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2364 of bytes you specify in the C<set_digest()> function (in this case 32).
2366 =head1 CIRCULAR REFERENCES
2368 DBM::Deep has B<experimental> support for circular references. Meaning you
2369 can have a nested hash key or array element that points to a parent object.
2370 This relationship is stored in the DB file, and is preserved between sessions.
2373 my $db = DBM::Deep->new( "foo.db" );
2376 $db->{circle} = $db; # ref to self
2378 print $db->{foo} . "\n"; # prints "foo"
2379 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2381 One catch is, passing the object to a function that recursively walks the
2382 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2383 C<export()> methods) will result in an infinite loop. The other catch is,
2384 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2385 or C<next_key()> methods), you will get the I<target object's key>, not the
2386 ref's key. This gets even more interesting with the above example, where
2387 the I<circle> key points to the base DB object, which technically doesn't
2388 have a key. So I made DBM::Deep return "[base]" as the key name in that
2391 =head1 CAVEATS / ISSUES / BUGS
2393 This section describes all the known issues with DBM::Deep. It you have found
2394 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2396 =head2 UNUSED SPACE RECOVERY
2398 One major caveat with DBM::Deep is that space occupied by existing keys and
2399 values is not recovered when they are deleted. Meaning if you keep deleting
2400 and adding new keys, your file will continuously grow. I am working on this,
2401 but in the meantime you can call the built-in C<optimize()> method from time to
2402 time (perhaps in a crontab or something) to recover all your unused space.
2404 $db->optimize(); # returns true on success
2406 This rebuilds the ENTIRE database into a new file, then moves it on top of
2407 the original. The new file will have no unused space, thus it will take up as
2408 little disk space as possible. Please note that this operation can take
2409 a long time for large files, and you need enough disk space to temporarily hold
2410 2 copies of your DB file. The temporary file is created in the same directory
2411 as the original, named with a ".tmp" extension, and is deleted when the
2412 operation completes. Oh, and if locking is enabled, the DB is automatically
2413 locked for the entire duration of the copy.
2415 B<WARNING:> Only call optimize() on the top-level node of the database, and
2416 make sure there are no child references lying around. DBM::Deep keeps a reference
2417 counter, and if it is greater than 1, optimize() will abort and return undef.
2419 =head2 AUTOVIVIFICATION
2421 Unfortunately, autovivification doesn't work with tied hashes. This appears to
2422 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
2423 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
2424 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
2425 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
2428 $db->{foo}->{bar} = "hello";
2430 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
2431 being an empty hash. Try this instead, which works fine:
2433 $db->{foo} = { bar => "hello" };
2435 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
2436 the execution path, and Perl indeed passes an empty hash to the STORE() method.
2437 Probably a bug in Perl.
2439 =head2 FILE CORRUPTION
2441 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2442 for a 32-bit signature when opened, but other corruption in files can cause
2443 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2444 stuck in an infinite loop depending on the level of corruption. File write
2445 operations are not checked for failure (for speed), so if you happen to run
2446 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2447 be addressed in a later version of DBM::Deep.
2451 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2452 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2453 about setting up your NFS server with a locking daemon, then using lockf() to
2454 lock your files, but your mileage may vary there as well. From what I
2455 understand, there is no real way to do it. However, if you need access to the
2456 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
2457 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2459 =head2 COPYING OBJECTS
2461 Beware of copying tied objects in Perl. Very strange things can happen.
2462 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2463 returns a new, blessed, tied hash or array to the same level in the DB.
2465 my $copy = $db->clone();
2467 B<Note>: Since clone() here is cloning the object, not the database location, any
2468 modifications to either $db or $copy will be visible in both.
2472 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2473 These functions cause every element in the array to move, which can be murder
2474 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2475 a different location. This will be addressed in the forthcoming version 1.00.
2477 =head2 WRITEONLY FILES
2479 If you pass in a filehandle to new(), you may have opened it in either a readonly or
2480 writeonly mode. STORE will verify that the filehandle is writable. However, there
2481 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
2482 filehandle isn't readable, it's not clear what will happen. So, don't do that.
2486 This section discusses DBM::Deep's speed and memory usage.
2490 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2491 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2492 multi-level hash/array support, and cross-platform FTPable files. Even so,
2493 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2494 with huge databases. Here is some test data:
2496 Adding 1,000,000 keys to new DB file...
2498 At 100 keys, avg. speed is 2,703 keys/sec
2499 At 200 keys, avg. speed is 2,642 keys/sec
2500 At 300 keys, avg. speed is 2,598 keys/sec
2501 At 400 keys, avg. speed is 2,578 keys/sec
2502 At 500 keys, avg. speed is 2,722 keys/sec
2503 At 600 keys, avg. speed is 2,628 keys/sec
2504 At 700 keys, avg. speed is 2,700 keys/sec
2505 At 800 keys, avg. speed is 2,607 keys/sec
2506 At 900 keys, avg. speed is 2,190 keys/sec
2507 At 1,000 keys, avg. speed is 2,570 keys/sec
2508 At 2,000 keys, avg. speed is 2,417 keys/sec
2509 At 3,000 keys, avg. speed is 1,982 keys/sec
2510 At 4,000 keys, avg. speed is 1,568 keys/sec
2511 At 5,000 keys, avg. speed is 1,533 keys/sec
2512 At 6,000 keys, avg. speed is 1,787 keys/sec
2513 At 7,000 keys, avg. speed is 1,977 keys/sec
2514 At 8,000 keys, avg. speed is 2,028 keys/sec
2515 At 9,000 keys, avg. speed is 2,077 keys/sec
2516 At 10,000 keys, avg. speed is 2,031 keys/sec
2517 At 20,000 keys, avg. speed is 1,970 keys/sec
2518 At 30,000 keys, avg. speed is 2,050 keys/sec
2519 At 40,000 keys, avg. speed is 2,073 keys/sec
2520 At 50,000 keys, avg. speed is 1,973 keys/sec
2521 At 60,000 keys, avg. speed is 1,914 keys/sec
2522 At 70,000 keys, avg. speed is 2,091 keys/sec
2523 At 80,000 keys, avg. speed is 2,103 keys/sec
2524 At 90,000 keys, avg. speed is 1,886 keys/sec
2525 At 100,000 keys, avg. speed is 1,970 keys/sec
2526 At 200,000 keys, avg. speed is 2,053 keys/sec
2527 At 300,000 keys, avg. speed is 1,697 keys/sec
2528 At 400,000 keys, avg. speed is 1,838 keys/sec
2529 At 500,000 keys, avg. speed is 1,941 keys/sec
2530 At 600,000 keys, avg. speed is 1,930 keys/sec
2531 At 700,000 keys, avg. speed is 1,735 keys/sec
2532 At 800,000 keys, avg. speed is 1,795 keys/sec
2533 At 900,000 keys, avg. speed is 1,221 keys/sec
2534 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2536 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2537 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2538 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2539 Run time was 12 min 3 sec.
2543 One of the great things about DBM::Deep is that it uses very little memory.
2544 Even with huge databases (1,000,000+ keys) you will not see much increased
2545 memory on your process. DBM::Deep relies solely on the filesystem for storing
2546 and fetching data. Here is output from I</usr/bin/top> before even opening a
2549 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2550 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2552 Basically the process is taking 2,716K of memory. And here is the same
2553 process after storing and fetching 1,000,000 keys:
2555 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2556 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2558 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2559 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2561 =head1 DB FILE FORMAT
2563 In case you were interested in the underlying DB file format, it is documented
2564 here in this section. You don't need to know this to use the module, it's just
2565 included for reference.
2569 DBM::Deep files always start with a 32-bit signature to identify the file type.
2570 This is at offset 0. The signature is "DPDB" in network byte order. This is
2571 checked for when the file is opened and an error will be thrown if it's not found.
2575 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2576 has a standard header containing the type of data, the length of data, and then
2577 the data itself. The type is a single character (1 byte), the length is a
2578 32-bit unsigned long in network byte order, and the data is, well, the data.
2579 Here is how it unfolds:
2583 Immediately after the 32-bit file signature is the I<Master Index> record.
2584 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2585 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2586 depending on how the DBM::Deep object was constructed.
2588 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2589 number). The first 8-bit char of the MD5 signature is the offset into the
2590 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2591 index element is a file offset of the next tag for the key/element in question,
2592 which is usually a I<Bucket List> tag (see below).
2594 The next tag I<could> be another index, depending on how many keys/elements
2595 exist. See L<RE-INDEXING> below for details.
2599 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2600 file offsets to where the actual data is stored. It starts with a standard
2601 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2602 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2603 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2604 When the list fills up, a I<Re-Index> operation is performed (See
2605 L<RE-INDEXING> below).
2609 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2610 index/value pair (in array mode). It starts with a standard tag header with
2611 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2612 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2613 header. The size reported in the tag header is only for the value, but then,
2614 just after the value is another size (32-bit unsigned long) and then the plain
2615 key itself. Since the value is likely to be fetched more often than the plain
2616 key, I figured it would be I<slightly> faster to store the value first.
2618 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2619 record for the nested structure, where the process begins all over again.
2623 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2624 exhausted. Then, when another key/element comes in, the list is converted to a
2625 new index record. However, this index will look at the next char in the MD5
2626 hash, and arrange new Bucket List pointers accordingly. This process is called
2627 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2628 17 (16 + new one) keys/elements are removed from the old Bucket List and
2629 inserted into the new index. Several new Bucket Lists are created in the
2630 process, as a new MD5 char from the key is being examined (it is unlikely that
2631 the keys will all share the same next char of their MD5s).
2633 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2634 when the Bucket Lists will turn into indexes, but the first round tends to
2635 happen right around 4,000 keys. You will see a I<slight> decrease in
2636 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2637 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2638 right around 900,000 keys. This process can continue nearly indefinitely --
2639 right up until the point the I<MD5> signatures start colliding with each other,
2640 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2641 getting struck by lightning while you are walking to cash in your tickets.
2642 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2643 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2644 this is 340 unodecillion, but don't quote me).
2648 When a new key/element is stored, the key (or index number) is first run through
2649 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2650 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2651 for the first char of the signature (in this case I<b0>). If it does not exist,
2652 a new I<Bucket List> is created for our key (and the next 15 future keys that
2653 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2654 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2655 this point, unless we are replacing an existing I<Bucket>), where the actual
2656 data will be stored.
2660 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2661 (or index number), then walking along the indexes. If there are enough
2662 keys/elements in this DB level, there might be nested indexes, each linked to
2663 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2664 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2665 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2666 plain key are stored.
2668 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2669 methods. In this process the indexes are walked systematically, and each key
2670 fetched in increasing MD5 order (which is why it appears random). Once the
2671 I<Bucket> is found, the value is skipped and the plain key returned instead.
2672 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2673 alphabetically sorted. This only happens on an index-level -- as soon as the
2674 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2675 so it's pretty much undefined how the keys will come out -- just like Perl's
2678 =head1 CODE COVERAGE
2680 We use B<Devel::Cover> to test the code coverage of our tests, below is the
2681 B<Devel::Cover> report on this module's test suite.
2683 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2684 File stmt bran cond sub pod time total
2685 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2686 blib/lib/DBM/Deep.pm 95.2 83.8 70.0 98.2 100.0 58.0 91.0
2687 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 n/a 26.7 98.0
2688 blib/lib/DBM/Deep/Hash.pm 95.3 80.0 100.0 100.0 n/a 15.3 92.4
2689 Total 96.2 84.8 74.4 98.8 100.0 100.0 92.4
2690 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2692 =head1 MORE INFORMATION
2694 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
2695 or send email to L<DBM-Deep@googlegroups.com>.
2699 Joseph Huckaby, L<jhuckaby@cpan.org>
2701 Rob Kinyon, L<rkinyon@cpan.org>
2703 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2707 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2708 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2712 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
2713 This is free software, you may use it and distribute it under the
2714 same terms as Perl itself.