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.
34 use Fcntl qw( :DEFAULT :flock :seek );
38 use vars qw( $VERSION );
42 # Set to 4 and 'N' for 32-bit offset tags (default). Theoretical limit of 4 GB per file.
43 # (Perl must be compiled with largefile support for files > 2 GB)
45 # Set to 8 and 'Q' for 64-bit offsets. Theoretical limit of 16 XB per file.
46 # (Perl must be compiled with largefile and 64-bit long support)
52 # Set to 4 and 'N' for 32-bit data length prefixes. Limit of 4 GB for each key/value.
53 # Upgrading this is possible (see above) but probably not necessary. If you need
54 # more than 4 GB for a single key or value, this module is really not for you :-)
56 #my $DATA_LENGTH_SIZE = 4;
57 #my $DATA_LENGTH_PACK = 'N';
58 our ($LONG_SIZE, $LONG_PACK, $DATA_LENGTH_SIZE, $DATA_LENGTH_PACK);
61 # Maximum number of buckets per list before another level of indexing is done.
62 # Increase this value for slightly greater speed, but larger database files.
63 # DO NOT decrease this value below 16, due to risk of recursive reindex overrun.
68 # Better not adjust anything below here, unless you're me :-)
72 # Setup digest function for keys
74 our ($DIGEST_FUNC, $HASH_SIZE);
75 #my $DIGEST_FUNC = \&Digest::MD5::md5;
78 # Precalculate index and bucket sizes based on values above.
81 my ($INDEX_SIZE, $BUCKET_SIZE, $BUCKET_LIST_SIZE);
88 # Setup file and tag signatures. These should never change.
90 sub SIG_FILE () { 'DPDB' }
91 sub SIG_HASH () { 'H' }
92 sub SIG_ARRAY () { 'A' }
93 sub SIG_SCALAR () { 'S' }
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 () { SIG_HASH }
104 sub TYPE_ARRAY () { SIG_ARRAY }
105 sub TYPE_SCALAR () { SIG_SCALAR }
111 if (scalar(@_) > 1) {
113 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
117 elsif ( ref $_[0] ) {
118 unless ( eval { local $SIG{'__DIE__'}; %{$_[0]} || 1 } ) {
119 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
124 $args = { file => shift };
132 # Class constructor method for Perl OO interface.
133 # Calls tie() and returns blessed reference to tied hash or array,
134 # providing a hybrid OO/tie interface.
137 my $args = $class->_get_args( @_ );
140 # Check if we want a tied hash or array.
143 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
144 $class = 'DBM::Deep::Array';
145 require DBM::Deep::Array;
146 tie @$self, $class, %$args;
149 $class = 'DBM::Deep::Hash';
150 require DBM::Deep::Hash;
151 tie %$self, $class, %$args;
154 return bless $self, $class;
159 # Setup $self and bless into this class.
164 # These are the defaults to be optionally overridden below
167 base_offset => length(SIG_FILE),
170 foreach my $param ( keys %$self ) {
171 next unless exists $args->{$param};
172 $self->{$param} = delete $args->{$param}
175 # locking implicitly enables autoflush
176 if ($args->{locking}) { $args->{autoflush} = 1; }
178 $self->{root} = exists $args->{root}
180 : DBM::Deep::_::Root->new( $args );
182 if (!defined($self->_fh)) { $self->_open(); }
189 require DBM::Deep::Hash;
190 return DBM::Deep::Hash->TIEHASH( @_ );
195 require DBM::Deep::Array;
196 return DBM::Deep::Array->TIEARRAY( @_ );
199 #XXX Unneeded now ...
205 # Open a fh to the database, create if nonexistent.
206 # Make sure file signature matches DBM::Deep spec.
208 my $self = $_[0]->_get_self;
210 if (defined($self->_fh)) { $self->_close(); }
213 local $SIG{'__DIE__'};
214 # Theoretically, adding O_BINARY should remove the need for the binmode
215 # Of course, testing it is going to be ... interesting.
216 my $flags = O_RDWR | O_CREAT | O_BINARY;
219 sysopen( $fh, $self->_root->{file}, $flags )
221 $self->_root->{fh} = $fh;
222 }; if ($@ ) { $self->_throw_error( "Received error: $@\n" ); }
223 if (! defined($self->_fh)) {
224 return $self->_throw_error("Cannot sysopen file: " . $self->_root->{file} . ": $!");
229 #XXX Can we remove this by using the right sysopen() flags?
230 # Maybe ... q.v. above
231 binmode $fh; # for win32
233 if ($self->_root->{autoflush}) {
234 my $old = select $fh;
240 seek($fh, 0, SEEK_SET);
243 my $bytes_read = read( $fh, $signature, length(SIG_FILE));
246 # File is empty -- write signature and master index
249 seek($fh, 0, SEEK_SET);
251 $self->_create_tag($self->_base_offset, $self->_type, chr(0) x $INDEX_SIZE);
253 my $plain_key = "[base]";
254 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
256 # Flush the filehandle
257 my $old_fh = select $fh;
263 my @stats = stat($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, $offset, SEEK_SET);
319 print($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, SEEK_SET);
343 if (eof $fh) { return undef; }
346 read( $fh, $sig, SIG_SIZE);
349 read( $fh, $size, $DATA_LENGTH_SIZE);
350 $size = unpack($DATA_LENGTH_PACK, $size);
353 read( $fh, $buffer, $size);
358 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
365 # Given index tag, lookup single entry in index and return .
368 my ($tag, $index) = @_;
370 my $location = unpack($LONG_PACK, substr($tag->{content}, $index * $LONG_SIZE, $LONG_SIZE) );
371 if (!$location) { return; }
373 return $self->_load_tag( $location );
378 # Adds one key/value pair to bucket list, given offset, MD5 digest of key,
379 # plain (undigested) key and value.
382 my ($tag, $md5, $plain_key, $value) = @_;
383 my $keys = $tag->{content};
387 # added ref() check first to avoid eval and runtime exception for every
388 # scalar value being stored. performance tweak.
389 my $is_dbm_deep = eval { local $SIG{'__DIE__'}; $value->isa( 'DBM::Deep' ) };
391 my $internal_ref = $is_dbm_deep && ($value->_root eq $self->_root);
396 # Iterate through buckets, seeing if this is a new entry or a replace.
398 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
399 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
400 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
403 # Found empty bucket (end of list). Populate and exit loop.
407 $location = $internal_ref
408 ? $value->_base_offset
409 : $self->_root->{end};
411 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
412 print($fh $md5 . pack($LONG_PACK, $location) );
415 elsif ($md5 eq $key) {
417 # Found existing bucket with same key. Replace with new value.
422 $location = $value->_base_offset;
423 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
424 print($fh $md5 . pack($LONG_PACK, $location) );
427 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
429 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
432 # If value is a hash, array, or raw value with equal or less size, we can
433 # reuse the same content area of the database. Otherwise, we have to create
434 # a new content area at the EOF.
437 my $r = Scalar::Util::reftype( $value ) || '';
438 if ( $r eq 'HASH' || $r eq 'ARRAY' ) {
439 $actual_length = $INDEX_SIZE;
441 # if autobless is enabled, must also take into consideration
442 # the class name, as it is stored along with key/value.
443 if ( $self->_root->{autobless} ) {
444 my $value_class = Scalar::Util::blessed($value);
445 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
446 $actual_length += length($value_class);
450 else { $actual_length = length($value); }
452 if ($actual_length <= $size) {
456 $location = $self->_root->{end};
457 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE, SEEK_SET);
458 print($fh pack($LONG_PACK, $location) );
466 # If this is an internal reference, return now.
467 # No need to write value or plain key
474 # If bucket didn't fit into list, split into a new index level
477 seek($fh, $tag->{ref_loc}, SEEK_SET);
478 print($fh pack($LONG_PACK, $self->_root->{end}) );
480 my $index_tag = $self->_create_tag($self->_root->{end}, SIG_INDEX, chr(0) x $INDEX_SIZE);
483 $keys .= $md5 . pack($LONG_PACK, 0);
485 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
486 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
488 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
489 my $num = ord(substr($key, $tag->{ch} + 1, 1));
491 if ($offsets[$num]) {
492 my $offset = $offsets[$num] + SIG_SIZE + $DATA_LENGTH_SIZE;
493 seek($fh, $offset, SEEK_SET);
495 read( $fh, $subkeys, $BUCKET_LIST_SIZE);
497 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
498 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
500 seek($fh, $offset + ($k * $BUCKET_SIZE), SEEK_SET);
501 print($fh $key . pack($LONG_PACK, $old_subloc || $self->_root->{end}) );
507 $offsets[$num] = $self->_root->{end};
508 seek($fh, $index_tag->{offset} + ($num * $LONG_SIZE), SEEK_SET);
509 print($fh pack($LONG_PACK, $self->_root->{end}) );
511 my $blist_tag = $self->_create_tag($self->_root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
513 seek($fh, $blist_tag->{offset}, SEEK_SET);
514 print($fh $key . pack($LONG_PACK, $old_subloc || $self->_root->{end}) );
519 $location ||= $self->_root->{end};
520 } # re-index bucket list
523 # Seek to content area and store signature, value and plaintext key
527 seek($fh, $location, SEEK_SET);
530 # Write signature based on content type, set content length and write actual value.
532 my $r = Scalar::Util::reftype($value) || '';
534 print($fh TYPE_HASH );
535 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
536 $content_length = $INDEX_SIZE;
538 elsif ($r eq 'ARRAY') {
539 print($fh TYPE_ARRAY );
540 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
541 $content_length = $INDEX_SIZE;
543 elsif (!defined($value)) {
544 print($fh SIG_NULL );
545 print($fh pack($DATA_LENGTH_PACK, 0) );
549 print($fh SIG_DATA );
550 print($fh pack($DATA_LENGTH_PACK, length($value)) . $value );
551 $content_length = length($value);
555 # Plain key is stored AFTER value, as keys are typically fetched less often.
557 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
560 # If value is blessed, preserve class name
562 if ( $self->_root->{autobless} ) {
563 my $value_class = Scalar::Util::blessed($value);
564 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
566 # Blessed ref -- will restore later
569 print($fh pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
570 $content_length += 1;
571 $content_length += $DATA_LENGTH_SIZE + length($value_class);
575 $content_length += 1;
580 # If this is a new content area, advance EOF counter
582 if ($location == $self->_root->{end}) {
583 $self->_root->{end} += SIG_SIZE;
584 $self->_root->{end} += $DATA_LENGTH_SIZE + $content_length;
585 $self->_root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
589 # If content is a hash or array, create new child DBM::Deep object and
590 # pass each key or element to it.
593 my $branch = DBM::Deep->new(
595 base_offset => $location,
596 root => $self->_root,
598 foreach my $key (keys %{$value}) {
599 $branch->STORE( $key, $value->{$key} );
602 elsif ($r eq 'ARRAY') {
603 my $branch = DBM::Deep->new(
605 base_offset => $location,
606 root => $self->_root,
609 foreach my $element (@{$value}) {
610 $branch->STORE( $index, $element );
618 return $self->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
621 sub _get_bucket_value {
623 # Fetch single value given tag and MD5 digested key.
626 my ($tag, $md5) = @_;
627 my $keys = $tag->{content};
632 # Iterate through buckets, looking for a key match
635 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
636 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
637 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
641 # Hit end of list, no match
646 if ( $md5 ne $key ) {
651 # Found match -- seek to offset and read signature
654 seek($fh, $subloc, SEEK_SET);
655 read( $fh, $signature, SIG_SIZE);
658 # If value is a hash or array, return new DBM::Deep object with correct offset
660 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
661 my $obj = DBM::Deep->new(
663 base_offset => $subloc,
667 if ($self->_root->{autobless}) {
669 # Skip over value and plain key to see if object needs
672 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
675 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
676 if ($size) { seek($fh, $size, SEEK_CUR); }
679 read( $fh, $bless_bit, 1);
680 if (ord($bless_bit)) {
682 # Yes, object needs to be re-blessed
685 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
686 if ($size) { read( $fh, $class_name, $size); }
687 if ($class_name) { $obj = bless( $obj, $class_name ); }
695 # Otherwise return actual value
697 elsif ($signature eq SIG_DATA) {
700 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
701 if ($size) { read( $fh, $value, $size); }
706 # Key exists, but content is null
716 # Delete single key/value pair given tag and MD5 digested key.
719 my ($tag, $md5) = @_;
720 my $keys = $tag->{content};
725 # Iterate through buckets, looking for a key match
728 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
729 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
730 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
734 # Hit end of list, no match
739 if ( $md5 ne $key ) {
744 # Matched key -- delete bucket and return
746 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
747 print($fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
748 print($fh chr(0) x $BUCKET_SIZE );
758 # Check existence of single key given tag and MD5 digested key.
761 my ($tag, $md5) = @_;
762 my $keys = $tag->{content};
765 # Iterate through buckets, looking for a key match
768 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
769 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
770 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
774 # Hit end of list, no match
779 if ( $md5 ne $key ) {
784 # Matched key -- return true
792 sub _find_bucket_list {
794 # Locate offset for bucket list, given digested key
800 # Locate offset for bucket list using digest index system
803 my $tag = $self->_load_tag($self->_base_offset);
804 if (!$tag) { return; }
806 while ($tag->{signature} ne SIG_BLIST) {
807 $tag = $self->_index_lookup($tag, ord(substr($md5, $ch, 1)));
808 if (!$tag) { return; }
815 sub _traverse_index {
817 # Scan index and recursively step into deeper levels, looking for next key.
819 my ($self, $offset, $ch, $force_return_next) = @_;
820 $force_return_next = undef unless $force_return_next;
822 my $tag = $self->_load_tag( $offset );
826 if ($tag->{signature} ne SIG_BLIST) {
827 my $content = $tag->{content};
829 if ($self->{return_next}) { $start = 0; }
830 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
832 for (my $index = $start; $index < 256; $index++) {
833 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
835 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
836 if (defined($result)) { return $result; }
840 $self->{return_next} = 1;
843 elsif ($tag->{signature} eq SIG_BLIST) {
844 my $keys = $tag->{content};
845 if ($force_return_next) { $self->{return_next} = 1; }
848 # Iterate through buckets, looking for a key match
850 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
851 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
852 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
856 # End of bucket list -- return to outer loop
858 $self->{return_next} = 1;
861 elsif ($key eq $self->{prev_md5}) {
863 # Located previous key -- return next one found
865 $self->{return_next} = 1;
868 elsif ($self->{return_next}) {
870 # Seek to bucket location and skip over signature
872 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
875 # Skip over value to get to plain key
878 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
879 if ($size) { seek($fh, $size, SEEK_CUR); }
882 # Read in plain key and return as scalar
885 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
886 if ($size) { read( $fh, $plain_key, $size); }
892 $self->{return_next} = 1;
893 } # tag is a bucket list
900 # Locate next key, given digested previous one
902 my $self = $_[0]->_get_self;
904 $self->{prev_md5} = $_[1] ? $_[1] : undef;
905 $self->{return_next} = 0;
908 # If the previous key was not specifed, start at the top and
909 # return the first one found.
911 if (!$self->{prev_md5}) {
912 $self->{prev_md5} = chr(0) x $HASH_SIZE;
913 $self->{return_next} = 1;
916 return $self->_traverse_index( $self->_base_offset, 0 );
921 # If db locking is set, flock() the db file. If called multiple
922 # times before unlock(), then the same number of unlocks() must
923 # be called before the lock is released.
925 my $self = $_[0]->_get_self;
927 $type = LOCK_EX unless defined $type;
929 if (!defined($self->_fh)) { return; }
931 if ($self->_root->{locking}) {
932 if (!$self->_root->{locked}) {
933 flock($self->_fh, $type);
935 # refresh end counter in case file has changed size
936 my @stats = stat($self->_root->{file});
937 $self->_root->{end} = $stats[7];
939 # double-check file inode, in case another process
940 # has optimize()d our file while we were waiting.
941 if ($stats[1] != $self->_root->{inode}) {
942 $self->_open(); # re-open
943 flock($self->_fh, $type); # re-lock
944 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
947 $self->_root->{locked}++;
957 # If db locking is set, unlock the db file. See note in lock()
958 # regarding calling lock() multiple times.
960 my $self = $_[0]->_get_self;
962 if (!defined($self->_fh)) { return; }
964 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
965 $self->_root->{locked}--;
966 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
974 #XXX These uses of ref() need verified
977 # Copy single level of keys or elements to new DB handle.
978 # Recurse for nested structures
980 my $self = $_[0]->_get_self;
983 if ($self->_type eq TYPE_HASH) {
984 my $key = $self->first_key();
986 my $value = $self->get($key);
987 #XXX This doesn't work with autobless
988 if (!ref($value)) { $db_temp->{$key} = $value; }
990 my $type = $value->_type;
991 if ($type eq TYPE_HASH) { $db_temp->{$key} = {}; }
992 else { $db_temp->{$key} = []; }
993 $value->_copy_node( $db_temp->{$key} );
995 $key = $self->next_key($key);
999 my $length = $self->length();
1000 for (my $index = 0; $index < $length; $index++) {
1001 my $value = $self->get($index);
1002 if (!ref($value)) { $db_temp->[$index] = $value; }
1003 #XXX NO tests for this code
1005 my $type = $value->_type;
1006 if ($type eq TYPE_HASH) { $db_temp->[$index] = {}; }
1007 else { $db_temp->[$index] = []; }
1008 $value->_copy_node( $db_temp->[$index] );
1016 # Recursively export into standard Perl hashes and arrays.
1018 my $self = $_[0]->_get_self;
1021 if ($self->_type eq TYPE_HASH) { $temp = {}; }
1022 elsif ($self->_type eq TYPE_ARRAY) { $temp = []; }
1025 $self->_copy_node( $temp );
1033 # Recursively import Perl hash/array structure
1035 #XXX This use of ref() seems to be ok
1036 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
1038 my $self = $_[0]->_get_self;
1041 #XXX This use of ref() seems to be ok
1042 if (!ref($struct)) {
1044 # struct is not a reference, so just import based on our type
1048 if ($self->_type eq TYPE_HASH) { $struct = {@_}; }
1049 elsif ($self->_type eq TYPE_ARRAY) { $struct = [@_]; }
1052 my $r = Scalar::Util::reftype($struct) || '';
1053 if ($r eq "HASH" && $self->_type eq TYPE_HASH) {
1054 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
1056 elsif ($r eq "ARRAY" && $self->_type eq TYPE_ARRAY) {
1057 $self->push( @$struct );
1060 return $self->_throw_error("Cannot import: type mismatch");
1068 # Rebuild entire database into new file, then move
1069 # it back on top of original.
1071 my $self = $_[0]->_get_self;
1073 #XXX Need to create a new test for this
1074 # if ($self->_root->{links} > 1) {
1075 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
1078 my $db_temp = DBM::Deep->new(
1079 file => $self->_root->{file} . '.tmp',
1080 type => $self->_type
1083 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
1087 $self->_copy_node( $db_temp );
1091 # Attempt to copy user, group and permissions over to new file
1093 my @stats = stat($self->_fh);
1094 my $perms = $stats[2] & 07777;
1095 my $uid = $stats[4];
1096 my $gid = $stats[5];
1097 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
1098 chmod( $perms, $self->_root->{file} . '.tmp' );
1100 # q.v. perlport for more information on this variable
1101 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
1103 # Potential race condition when optmizing on Win32 with locking.
1104 # The Windows filesystem requires that the filehandle be closed
1105 # before it is overwritten with rename(). This could be redone
1112 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
1113 unlink $self->_root->{file} . '.tmp';
1115 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
1127 # Make copy of object and return
1129 my $self = $_[0]->_get_self;
1131 return DBM::Deep->new(
1132 type => $self->_type,
1133 base_offset => $self->_base_offset,
1134 root => $self->_root
1139 my %is_legal_filter = map {
1142 store_key store_value
1143 fetch_key fetch_value
1148 # Setup filter function for storing or fetching the key or value
1150 my $self = $_[0]->_get_self;
1151 my $type = lc $_[1];
1152 my $func = $_[2] ? $_[2] : undef;
1154 if ( $is_legal_filter{$type} ) {
1155 $self->_root->{"filter_$type"} = $func;
1169 # Get access to the root structure
1171 my $self = $_[0]->_get_self;
1172 return $self->{root};
1177 # Get access to the raw fh
1179 #XXX It will be useful, though, when we split out HASH and ARRAY
1180 my $self = $_[0]->_get_self;
1181 return $self->_root->{fh};
1186 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
1188 my $self = $_[0]->_get_self;
1189 return $self->{type};
1194 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
1196 my $self = $_[0]->_get_self;
1197 return $self->{base_offset};
1202 # Get last error string, or undef if no error
1205 #? ( _get_self($_[0])->{root}->{error} or undef )
1206 ? ( $_[0]->_get_self->{root}->{error} or undef )
1216 # Store error string in self
1218 my $error_text = $_[1];
1220 if ( Scalar::Util::blessed $_[0] ) {
1221 my $self = $_[0]->_get_self;
1222 $self->_root->{error} = $error_text;
1224 unless ($self->_root->{debug}) {
1225 die "DBM::Deep: $error_text\n";
1228 warn "DBM::Deep: $error_text\n";
1232 die "DBM::Deep: $error_text\n";
1240 my $self = $_[0]->_get_self;
1242 undef $self->_root->{error};
1245 sub _precalc_sizes {
1247 # Precalculate index, bucket and bucket list sizes
1250 #XXX I don't like this ...
1251 set_pack() unless defined $LONG_SIZE;
1253 $INDEX_SIZE = 256 * $LONG_SIZE;
1254 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
1255 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
1260 # Set pack/unpack modes (see file header for more)
1262 my ($long_s, $long_p, $data_s, $data_p) = @_;
1264 $LONG_SIZE = $long_s ? $long_s : 4;
1265 $LONG_PACK = $long_p ? $long_p : 'N';
1267 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
1268 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
1275 # Set key digest function (default is MD5)
1277 my ($digest_func, $hash_size) = @_;
1279 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
1280 $HASH_SIZE = $hash_size ? $hash_size : 16;
1286 # tie() methods (hashes and arrays)
1291 # Store single hash key/value or array element in database.
1293 my $self = $_[0]->_get_self;
1296 # User may be storing a hash, in which case we do not want it run
1297 # through the filtering system
1298 my $value = ($self->_root->{filter_store_value} && !ref($_[2]))
1299 ? $self->_root->{filter_store_value}->($_[2])
1302 my $md5 = $DIGEST_FUNC->($key);
1305 # Make sure file is open
1307 if (!defined($self->_fh) && !$self->_open()) {
1313 # Request exclusive lock for writing
1315 $self->lock( LOCK_EX );
1317 my $fh = $self->_fh;
1320 # Locate offset for bucket list using digest index system
1322 my $tag = $self->_load_tag($self->_base_offset);
1324 $tag = $self->_create_tag($self->_base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1328 while ($tag->{signature} ne SIG_BLIST) {
1329 my $num = ord(substr($md5, $ch, 1));
1330 my $new_tag = $self->_index_lookup($tag, $num);
1332 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1333 seek($fh, $ref_loc, SEEK_SET);
1334 print($fh pack($LONG_PACK, $self->_root->{end}) );
1336 $tag = $self->_create_tag($self->_root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1337 $tag->{ref_loc} = $ref_loc;
1342 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1344 $tag->{ref_loc} = $ref_loc;
1351 # Add key/value to bucket list
1353 my $result = $self->_add_bucket( $tag, $md5, $key, $value );
1362 # Fetch single value or element given plain key or array index
1364 my $self = shift->_get_self;
1368 # Make sure file is open
1370 if (!defined($self->_fh)) { $self->_open(); }
1372 my $md5 = $DIGEST_FUNC->($key);
1375 # Request shared lock for reading
1377 $self->lock( LOCK_SH );
1379 my $tag = $self->_find_bucket_list( $md5 );
1386 # Get value from bucket list
1388 my $result = $self->_get_bucket_value( $tag, $md5 );
1392 #XXX What is ref() checking here?
1393 #YYY Filters only apply on scalar values, so the ref check is making
1394 #YYY sure the fetched bucket is a scalar, not a child hash or array.
1395 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
1396 ? $self->_root->{filter_fetch_value}->($result)
1402 # Delete single key/value pair or element given plain key or array index
1404 my $self = $_[0]->_get_self;
1407 my $md5 = $DIGEST_FUNC->($key);
1410 # Make sure file is open
1412 if (!defined($self->_fh)) { $self->_open(); }
1415 # Request exclusive lock for writing
1417 $self->lock( LOCK_EX );
1419 my $tag = $self->_find_bucket_list( $md5 );
1428 my $value = $self->_get_bucket_value( $tag, $md5 );
1429 if ($value && !ref($value) && $self->_root->{filter_fetch_value}) {
1430 $value = $self->_root->{filter_fetch_value}->($value);
1433 my $result = $self->_delete_bucket( $tag, $md5 );
1436 # If this object is an array and the key deleted was on the end of the stack,
1437 # decrement the length variable.
1447 # Check if a single key or element exists given plain key or array index
1449 my $self = $_[0]->_get_self;
1452 my $md5 = $DIGEST_FUNC->($key);
1455 # Make sure file is open
1457 if (!defined($self->_fh)) { $self->_open(); }
1460 # Request shared lock for reading
1462 $self->lock( LOCK_SH );
1464 my $tag = $self->_find_bucket_list( $md5 );
1467 # For some reason, the built-in exists() function returns '' for false
1475 # Check if bucket exists and return 1 or ''
1477 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1486 # Clear all keys from hash, or all elements from array.
1488 my $self = $_[0]->_get_self;
1491 # Make sure file is open
1493 if (!defined($self->_fh)) { $self->_open(); }
1496 # Request exclusive lock for writing
1498 $self->lock( LOCK_EX );
1500 my $fh = $self->_fh;
1502 seek($fh, $self->_base_offset, SEEK_SET);
1508 $self->_create_tag($self->_base_offset, $self->_type, chr(0) x $INDEX_SIZE);
1516 # Public method aliases
1518 sub put { (shift)->STORE( @_ ) }
1519 sub store { (shift)->STORE( @_ ) }
1520 sub get { (shift)->FETCH( @_ ) }
1521 sub fetch { (shift)->FETCH( @_ ) }
1522 sub delete { (shift)->DELETE( @_ ) }
1523 sub exists { (shift)->EXISTS( @_ ) }
1524 sub clear { (shift)->CLEAR( @_ ) }
1526 package DBM::Deep::_::Root;
1539 filter_store_key => undef,
1540 filter_store_value => undef,
1541 filter_fetch_key => undef,
1542 filter_fetch_value => undef,
1553 return unless $self;
1555 close $self->{fh} if $self->{fh};
1566 DBM::Deep - A pure perl multi-level hash/array DBM
1571 my $db = DBM::Deep->new( "foo.db" );
1573 $db->{key} = 'value'; # tie() style
1576 $db->put('key' => 'value'); # OO style
1577 print $db->get('key');
1579 # true multi-level support
1580 $db->{my_complex} = [
1581 'hello', { perl => 'rules' },
1587 A unique flat-file database module, written in pure perl. True
1588 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1589 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1590 handle millions of keys and unlimited hash levels without significant
1591 slow-down. Written from the ground-up in pure perl -- this is NOT a
1592 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1593 Mac OS X and Windows.
1597 Hopefully you are using Perl's excellent CPAN module, which will download
1598 and install the module for you. If not, get the tarball, and run these
1610 Construction can be done OO-style (which is the recommended way), or using
1611 Perl's tie() function. Both are examined here.
1613 =head2 OO CONSTRUCTION
1615 The recommended way to construct a DBM::Deep object is to use the new()
1616 method, which gets you a blessed, tied hash or array reference.
1618 my $db = DBM::Deep->new( "foo.db" );
1620 This opens a new database handle, mapped to the file "foo.db". If this
1621 file does not exist, it will automatically be created. DB files are
1622 opened in "r+" (read/write) mode, and the type of object returned is a
1623 hash, unless otherwise specified (see L<OPTIONS> below).
1625 You can pass a number of options to the constructor to specify things like
1626 locking, autoflush, etc. This is done by passing an inline hash:
1628 my $db = DBM::Deep->new(
1634 Notice that the filename is now specified I<inside> the hash with
1635 the "file" parameter, as opposed to being the sole argument to the
1636 constructor. This is required if any options are specified.
1637 See L<OPTIONS> below for the complete list.
1641 You can also start with an array instead of a hash. For this, you must
1642 specify the C<type> parameter:
1644 my $db = DBM::Deep->new(
1646 type => DBM::Deep->TYPE_ARRAY
1649 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1650 a new DB file. If you create a DBM::Deep object with an existing file, the
1651 C<type> will be loaded from the file header, and an error will be thrown if
1652 the wrong type is passed in.
1654 =head2 TIE CONSTRUCTION
1656 Alternately, you can create a DBM::Deep handle by using Perl's built-in
1657 tie() function. The object returned from tie() can be used to call methods,
1658 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
1659 file (as expected with most tie'd objects).
1662 my $db = tie %hash, "DBM::Deep", "foo.db";
1665 my $db = tie @array, "DBM::Deep", "bar.db";
1667 As with the OO constructor, you can replace the DB filename parameter with
1668 a hash containing one or more options (see L<OPTIONS> just below for the
1671 tie %hash, "DBM::Deep", {
1679 There are a number of options that can be passed in when constructing your
1680 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1686 Filename of the DB file to link the handle to. You can pass a full absolute
1687 filesystem path, partial path, or a plain filename if the file is in the
1688 current working directory. This is a required parameter.
1692 This parameter specifies what type of object to create, a hash or array. Use
1693 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1694 This only takes effect when beginning a new file. This is an optional
1695 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1699 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1700 function to lock the database in exclusive mode for writes, and shared mode for
1701 reads. Pass any true value to enable. This affects the base DB handle I<and
1702 any child hashes or arrays> that use the same DB file. This is an optional
1703 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1707 Specifies whether autoflush is to be enabled on the underlying filehandle.
1708 This obviously slows down write operations, but is required if you may have
1709 multiple processes accessing the same DB file (also consider enable I<locking>).
1710 Pass any true value to enable. This is an optional parameter, and defaults to 0
1715 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1716 restore them when fetched. This is an B<experimental> feature, and does have
1717 side-effects. Basically, when hashes are re-blessed into their original
1718 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1719 able to call any DBM::Deep methods on them. You have been warned.
1720 This is an optional parameter, and defaults to 0 (disabled).
1724 See L<FILTERS> below.
1728 Setting I<debug> mode will make all errors non-fatal, dump them out to
1729 STDERR, and continue on. This is for debugging purposes only, and probably
1730 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1734 Instead of passing a file path, you can instead pass a handle to an pre-opened
1735 filehandle. Note: Beware of using the magick *DATA handle, as this actually
1736 contains your entire Perl script, as well as the data following the __DATA__
1737 marker. This will not work, because DBM::Deep uses absolute seek()s into the
1738 file. Instead, consider reading *DATA into an IO::Scalar handle, then passing
1739 in that. Also please note optimize() will NOT work when passing in only a
1740 handle. Pass in a real filename in order to use optimize().
1744 =head1 TIE INTERFACE
1746 With DBM::Deep you can access your databases using Perl's standard hash/array
1747 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
1748 treat them as such. DBM::Deep will intercept all reads/writes and direct them
1749 to the right place -- the DB file. This has nothing to do with the
1750 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
1751 using regular hashes and arrays, rather than calling functions like C<get()>
1752 and C<put()> (although those work too). It is entirely up to you how to want
1753 to access your databases.
1757 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1758 or even nested hashes (or arrays) using standard Perl syntax:
1760 my $db = DBM::Deep->new( "foo.db" );
1762 $db->{mykey} = "myvalue";
1764 $db->{myhash}->{subkey} = "subvalue";
1766 print $db->{myhash}->{subkey} . "\n";
1768 You can even step through hash keys using the normal Perl C<keys()> function:
1770 foreach my $key (keys %$db) {
1771 print "$key: " . $db->{$key} . "\n";
1774 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1775 pushes them onto an array, all before the loop even begins. If you have an
1776 extra large hash, this may exhaust Perl's memory. Instead, consider using
1777 Perl's C<each()> function, which pulls keys/values one at a time, using very
1780 while (my ($key, $value) = each %$db) {
1781 print "$key: $value\n";
1784 Please note that when using C<each()>, you should always pass a direct
1785 hash reference, not a lookup. Meaning, you should B<never> do this:
1788 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1790 This causes an infinite loop, because for each iteration, Perl is calling
1791 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1792 it effectively keeps returning the first key over and over again. Instead,
1793 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1797 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1798 reference. This includes inserting, removing and manipulating elements,
1799 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1800 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1801 or simply be a nested array reference inside a hash. Example:
1803 my $db = DBM::Deep->new(
1804 file => "foo-array.db",
1805 type => DBM::Deep->TYPE_ARRAY
1809 push @$db, "bar", "baz";
1810 unshift @$db, "bah";
1812 my $last_elem = pop @$db; # baz
1813 my $first_elem = shift @$db; # bah
1814 my $second_elem = $db->[1]; # bar
1816 my $num_elements = scalar @$db;
1820 In addition to the I<tie()> interface, you can also use a standard OO interface
1821 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1822 array) has its own methods, but both types share the following common methods:
1823 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1827 =item * new() / clone()
1829 These are the constructor and copy-functions.
1831 =item * put() / store()
1833 Stores a new hash key/value pair, or sets an array element value. Takes two
1834 arguments, the hash key or array index, and the new value. The value can be
1835 a scalar, hash ref or array ref. Returns true on success, false on failure.
1837 $db->put("foo", "bar"); # for hashes
1838 $db->put(1, "bar"); # for arrays
1840 =item * get() / fetch()
1842 Fetches the value of a hash key or array element. Takes one argument: the hash
1843 key or array index. Returns a scalar, hash ref or array ref, depending on the
1846 my $value = $db->get("foo"); # for hashes
1847 my $value = $db->get(1); # for arrays
1851 Checks if a hash key or array index exists. Takes one argument: the hash key
1852 or array index. Returns true if it exists, false if not.
1854 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1855 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1859 Deletes one hash key/value pair or array element. Takes one argument: the hash
1860 key or array index. Returns true on success, false if not found. For arrays,
1861 the remaining elements located after the deleted element are NOT moved over.
1862 The deleted element is essentially just undefined, which is exactly how Perl's
1863 internal arrays work. Please note that the space occupied by the deleted
1864 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1865 below for details and workarounds.
1867 $db->delete("foo"); # for hashes
1868 $db->delete(1); # for arrays
1872 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1873 value. Please note that the space occupied by the deleted keys/values or
1874 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1875 details and workarounds.
1877 $db->clear(); # hashes or arrays
1879 =item * lock() / unlock()
1885 Recover lost disk space.
1887 =item * import() / export()
1889 Data going in and out.
1891 =item * set_digest() / set_pack() / set_filter()
1893 q.v. adjusting the interal parameters.
1895 =item * error() / clear_error()
1897 Error handling methods (may be deprecated).
1903 For hashes, DBM::Deep supports all the common methods described above, and the
1904 following additional methods: C<first_key()> and C<next_key()>.
1910 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1911 fetched in an undefined order (which appears random). Takes no arguments,
1912 returns the key as a scalar value.
1914 my $key = $db->first_key();
1918 Returns the "next" key in the hash, given the previous one as the sole argument.
1919 Returns undef if there are no more keys to be fetched.
1921 $key = $db->next_key($key);
1925 Here are some examples of using hashes:
1927 my $db = DBM::Deep->new( "foo.db" );
1929 $db->put("foo", "bar");
1930 print "foo: " . $db->get("foo") . "\n";
1932 $db->put("baz", {}); # new child hash ref
1933 $db->get("baz")->put("buz", "biz");
1934 print "buz: " . $db->get("baz")->get("buz") . "\n";
1936 my $key = $db->first_key();
1938 print "$key: " . $db->get($key) . "\n";
1939 $key = $db->next_key($key);
1942 if ($db->exists("foo")) { $db->delete("foo"); }
1946 For arrays, DBM::Deep supports all the common methods described above, and the
1947 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1948 C<unshift()> and C<splice()>.
1954 Returns the number of elements in the array. Takes no arguments.
1956 my $len = $db->length();
1960 Adds one or more elements onto the end of the array. Accepts scalars, hash
1961 refs or array refs. No return value.
1963 $db->push("foo", "bar", {});
1967 Fetches the last element in the array, and deletes it. Takes no arguments.
1968 Returns undef if array is empty. Returns the element value.
1970 my $elem = $db->pop();
1974 Fetches the first element in the array, deletes it, then shifts all the
1975 remaining elements over to take up the space. Returns the element value. This
1976 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1979 my $elem = $db->shift();
1983 Inserts one or more elements onto the beginning of the array, shifting all
1984 existing elements over to make room. Accepts scalars, hash refs or array refs.
1985 No return value. This method is not recommended with large arrays -- see
1986 <LARGE ARRAYS> below for details.
1988 $db->unshift("foo", "bar", {});
1992 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1993 -f splice> for usage -- it is too complicated to document here. This method is
1994 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1998 Here are some examples of using arrays:
2000 my $db = DBM::Deep->new(
2002 type => DBM::Deep->TYPE_ARRAY
2005 $db->push("bar", "baz");
2006 $db->unshift("foo");
2009 my $len = $db->length();
2010 print "length: $len\n"; # 4
2012 for (my $k=0; $k<$len; $k++) {
2013 print "$k: " . $db->get($k) . "\n";
2016 $db->splice(1, 2, "biz", "baf");
2018 while (my $elem = shift @$db) {
2019 print "shifted: $elem\n";
2024 Enable automatic file locking by passing a true value to the C<locking>
2025 parameter when constructing your DBM::Deep object (see L<SETUP> above).
2027 my $db = DBM::Deep->new(
2032 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
2033 mode for writes, and shared mode for reads. This is required if you have
2034 multiple processes accessing the same database file, to avoid file corruption.
2035 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
2036 NFS> below for more.
2038 =head2 EXPLICIT LOCKING
2040 You can explicitly lock a database, so it remains locked for multiple
2041 transactions. This is done by calling the C<lock()> method, and passing an
2042 optional lock mode argument (defaults to exclusive mode). This is particularly
2043 useful for things like counters, where the current value needs to be fetched,
2044 then incremented, then stored again.
2047 my $counter = $db->get("counter");
2049 $db->put("counter", $counter);
2058 You can pass C<lock()> an optional argument, which specifies which mode to use
2059 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2060 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2061 same as the constants defined in Perl's C<Fcntl> module.
2063 $db->lock( DBM::Deep->LOCK_SH );
2067 =head1 IMPORTING/EXPORTING
2069 You can import existing complex structures by calling the C<import()> method,
2070 and export an entire database into an in-memory structure using the C<export()>
2071 method. Both are examined here.
2075 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2076 walking the structure and adding keys/elements to the database as you go,
2077 simply pass a reference to the C<import()> method. This recursively adds
2078 everything to an existing DBM::Deep object for you. Here is an example:
2083 array1 => [ "elem0", "elem1", "elem2" ],
2085 subkey1 => "subvalue1",
2086 subkey2 => "subvalue2"
2090 my $db = DBM::Deep->new( "foo.db" );
2091 $db->import( $struct );
2093 print $db->{key1} . "\n"; # prints "value1"
2095 This recursively imports the entire C<$struct> object into C<$db>, including
2096 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2097 keys are merged with the existing ones, replacing if they already exist.
2098 The C<import()> method can be called on any database level (not just the base
2099 level), and works with both hash and array DB types.
2101 B<Note:> Make sure your existing structure has no circular references in it.
2102 These will cause an infinite loop when importing.
2106 Calling the C<export()> method on an existing DBM::Deep object will return
2107 a reference to a new in-memory copy of the database. The export is done
2108 recursively, so all nested hashes/arrays are all exported to standard Perl
2109 objects. Here is an example:
2111 my $db = DBM::Deep->new( "foo.db" );
2113 $db->{key1} = "value1";
2114 $db->{key2} = "value2";
2116 $db->{hash1}->{subkey1} = "subvalue1";
2117 $db->{hash1}->{subkey2} = "subvalue2";
2119 my $struct = $db->export();
2121 print $struct->{key1} . "\n"; # prints "value1"
2123 This makes a complete copy of the database in memory, and returns a reference
2124 to it. The C<export()> method can be called on any database level (not just
2125 the base level), and works with both hash and array DB types. Be careful of
2126 large databases -- you can store a lot more data in a DBM::Deep object than an
2127 in-memory Perl structure.
2129 B<Note:> Make sure your database has no circular references in it.
2130 These will cause an infinite loop when exporting.
2134 DBM::Deep has a number of hooks where you can specify your own Perl function
2135 to perform filtering on incoming or outgoing data. This is a perfect
2136 way to extend the engine, and implement things like real-time compression or
2137 encryption. Filtering applies to the base DB level, and all child hashes /
2138 arrays. Filter hooks can be specified when your DBM::Deep object is first
2139 constructed, or by calling the C<set_filter()> method at any time. There are
2140 four available filter hooks, described below:
2144 =item * filter_store_key
2146 This filter is called whenever a hash key is stored. It
2147 is passed the incoming key, and expected to return a transformed key.
2149 =item * filter_store_value
2151 This filter is called whenever a hash key or array element is stored. It
2152 is passed the incoming value, and expected to return a transformed value.
2154 =item * filter_fetch_key
2156 This filter is called whenever a hash key is fetched (i.e. via
2157 C<first_key()> or C<next_key()>). It is passed the transformed key,
2158 and expected to return the plain key.
2160 =item * filter_fetch_value
2162 This filter is called whenever a hash key or array element is fetched.
2163 It is passed the transformed value, and expected to return the plain value.
2167 Here are the two ways to setup a filter hook:
2169 my $db = DBM::Deep->new(
2171 filter_store_value => \&my_filter_store,
2172 filter_fetch_value => \&my_filter_fetch
2177 $db->set_filter( "filter_store_value", \&my_filter_store );
2178 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2180 Your filter function will be called only when dealing with SCALAR keys or
2181 values. When nested hashes and arrays are being stored/fetched, filtering
2182 is bypassed. Filters are called as static functions, passed a single SCALAR
2183 argument, and expected to return a single SCALAR value. If you want to
2184 remove a filter, set the function reference to C<undef>:
2186 $db->set_filter( "filter_store_value", undef );
2188 =head2 REAL-TIME ENCRYPTION EXAMPLE
2190 Here is a working example that uses the I<Crypt::Blowfish> module to
2191 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2192 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2193 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2196 use Crypt::Blowfish;
2199 my $cipher = Crypt::CBC->new({
2200 'key' => 'my secret key',
2201 'cipher' => 'Blowfish',
2203 'regenerate_key' => 0,
2204 'padding' => 'space',
2208 my $db = DBM::Deep->new(
2209 file => "foo-encrypt.db",
2210 filter_store_key => \&my_encrypt,
2211 filter_store_value => \&my_encrypt,
2212 filter_fetch_key => \&my_decrypt,
2213 filter_fetch_value => \&my_decrypt,
2216 $db->{key1} = "value1";
2217 $db->{key2} = "value2";
2218 print "key1: " . $db->{key1} . "\n";
2219 print "key2: " . $db->{key2} . "\n";
2225 return $cipher->encrypt( $_[0] );
2228 return $cipher->decrypt( $_[0] );
2231 =head2 REAL-TIME COMPRESSION EXAMPLE
2233 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2234 compression / decompression of keys & values with DBM::Deep Filters.
2235 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2236 more on I<Compress::Zlib>.
2241 my $db = DBM::Deep->new(
2242 file => "foo-compress.db",
2243 filter_store_key => \&my_compress,
2244 filter_store_value => \&my_compress,
2245 filter_fetch_key => \&my_decompress,
2246 filter_fetch_value => \&my_decompress,
2249 $db->{key1} = "value1";
2250 $db->{key2} = "value2";
2251 print "key1: " . $db->{key1} . "\n";
2252 print "key2: " . $db->{key2} . "\n";
2258 return Compress::Zlib::memGzip( $_[0] ) ;
2261 return Compress::Zlib::memGunzip( $_[0] ) ;
2264 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2265 actually numerical index numbers, and are not filtered.
2267 =head1 ERROR HANDLING
2269 Most DBM::Deep methods return a true value for success, and call die() on
2270 failure. You can wrap calls in an eval block to catch the die. Also, the
2271 actual error message is stored in an internal scalar, which can be fetched by
2272 calling the C<error()> method.
2274 my $db = DBM::Deep->new( "foo.db" ); # create hash
2275 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2277 print $@; # prints error message
2278 print $db->error(); # prints error message
2280 You can then call C<clear_error()> to clear the current error state.
2284 If you set the C<debug> option to true when creating your DBM::Deep object,
2285 all errors are considered NON-FATAL, and dumped to STDERR. This should only
2286 be used for debugging purposes and not production work. DBM::Deep expects errors
2287 to be thrown, not propagated back up the stack.
2289 =head1 LARGEFILE SUPPORT
2291 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2292 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2293 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2294 by calling the static C<set_pack()> method before you do anything else.
2296 DBM::Deep::set_pack(8, 'Q');
2298 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2299 instead of 32-bit longs. After setting these values your DB files have a
2300 theoretical maximum size of 16 XB (exabytes).
2302 B<Note:> Changing these values will B<NOT> work for existing database files.
2303 Only change this for new files, and make sure it stays set consistently
2304 throughout the file's life. If you do set these values, you can no longer
2305 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2306 back to 32-bit mode.
2308 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2309 only a 32-bit Perl. However, I have received user reports that this does
2312 =head1 LOW-LEVEL ACCESS
2314 If you require low-level access to the underlying filehandle that DBM::Deep uses,
2315 you can call the C<_fh()> method, which returns the handle:
2317 my $fh = $db->_fh();
2319 This method can be called on the root level of the datbase, or any child
2320 hashes or arrays. All levels share a I<root> structure, which contains things
2321 like the filehandle, a reference counter, and all the options specified
2322 when you created the object. You can get access to this root structure by
2323 calling the C<root()> method.
2325 my $root = $db->_root();
2327 This is useful for changing options after the object has already been created,
2328 such as enabling/disabling locking, or debug modes. You can also
2329 store your own temporary user data in this structure (be wary of name
2330 collision), which is then accessible from any child hash or array.
2332 =head1 CUSTOM DIGEST ALGORITHM
2334 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2335 keys. However you can override this, and use another algorithm (such as SHA-256)
2336 or even write your own. But please note that DBM::Deep currently expects zero
2337 collisions, so your algorithm has to be I<perfect>, so to speak.
2338 Collision detection may be introduced in a later version.
2342 You can specify a custom digest algorithm by calling the static C<set_digest()>
2343 function, passing a reference to a subroutine, and the length of the algorithm's
2344 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2345 objects. Here is a working example that uses a 256-bit hash from the
2346 I<Digest::SHA256> module. Please see
2347 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2352 my $context = Digest::SHA256::new(256);
2354 DBM::Deep::set_digest( \&my_digest, 32 );
2356 my $db = DBM::Deep->new( "foo-sha.db" );
2358 $db->{key1} = "value1";
2359 $db->{key2} = "value2";
2360 print "key1: " . $db->{key1} . "\n";
2361 print "key2: " . $db->{key2} . "\n";
2367 return substr( $context->hash($_[0]), 0, 32 );
2370 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2371 of bytes you specify in the C<set_digest()> function (in this case 32).
2373 =head1 CIRCULAR REFERENCES
2375 DBM::Deep has B<experimental> support for circular references. Meaning you
2376 can have a nested hash key or array element that points to a parent object.
2377 This relationship is stored in the DB file, and is preserved between sessions.
2380 my $db = DBM::Deep->new( "foo.db" );
2383 $db->{circle} = $db; # ref to self
2385 print $db->{foo} . "\n"; # prints "foo"
2386 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2388 One catch is, passing the object to a function that recursively walks the
2389 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2390 C<export()> methods) will result in an infinite loop. The other catch is,
2391 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2392 or C<next_key()> methods), you will get the I<target object's key>, not the
2393 ref's key. This gets even more interesting with the above example, where
2394 the I<circle> key points to the base DB object, which technically doesn't
2395 have a key. So I made DBM::Deep return "[base]" as the key name in that
2398 =head1 CAVEATS / ISSUES / BUGS
2400 This section describes all the known issues with DBM::Deep. It you have found
2401 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2403 =head2 UNUSED SPACE RECOVERY
2405 One major caveat with DBM::Deep is that space occupied by existing keys and
2406 values is not recovered when they are deleted. Meaning if you keep deleting
2407 and adding new keys, your file will continuously grow. I am working on this,
2408 but in the meantime you can call the built-in C<optimize()> method from time to
2409 time (perhaps in a crontab or something) to recover all your unused space.
2411 $db->optimize(); # returns true on success
2413 This rebuilds the ENTIRE database into a new file, then moves it on top of
2414 the original. The new file will have no unused space, thus it will take up as
2415 little disk space as possible. Please note that this operation can take
2416 a long time for large files, and you need enough disk space to temporarily hold
2417 2 copies of your DB file. The temporary file is created in the same directory
2418 as the original, named with a ".tmp" extension, and is deleted when the
2419 operation completes. Oh, and if locking is enabled, the DB is automatically
2420 locked for the entire duration of the copy.
2422 B<WARNING:> Only call optimize() on the top-level node of the database, and
2423 make sure there are no child references lying around. DBM::Deep keeps a reference
2424 counter, and if it is greater than 1, optimize() will abort and return undef.
2426 =head2 AUTOVIVIFICATION
2428 Unfortunately, autovivification doesn't work with tied hashes. This appears to
2429 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
2430 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
2431 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
2432 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
2435 $db->{foo}->{bar} = "hello";
2437 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
2438 being an empty hash. Try this instead, which works fine:
2440 $db->{foo} = { bar => "hello" };
2442 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
2443 the execution path, and Perl indeed passes an empty hash to the STORE() method.
2444 Probably a bug in Perl.
2446 =head2 FILE CORRUPTION
2448 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2449 for a 32-bit signature when opened, but other corruption in files can cause
2450 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2451 stuck in an infinite loop depending on the level of corruption. File write
2452 operations are not checked for failure (for speed), so if you happen to run
2453 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2454 be addressed in a later version of DBM::Deep.
2458 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2459 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2460 about setting up your NFS server with a locking daemon, then using lockf() to
2461 lock your files, but your mileage may vary there as well. From what I
2462 understand, there is no real way to do it. However, if you need access to the
2463 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
2464 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2466 =head2 COPYING OBJECTS
2468 Beware of copying tied objects in Perl. Very strange things can happen.
2469 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2470 returns a new, blessed, tied hash or array to the same level in the DB.
2472 my $copy = $db->clone();
2474 B<Note>: Since clone() here is cloning the object, not the database location, any
2475 modifications to either $db or $copy will be visible in both.
2479 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2480 These functions cause every element in the array to move, which can be murder
2481 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2482 a different location. This will be addressed in the forthcoming version 1.00.
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 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 93.9 82.5 70.0 96.5 33.3 84.3 88.1
2687 blib/lib/DBM/Deep/Array.pm 98.8 88.9 87.5 100.0 n/a 9.0 96.4
2688 blib/lib/DBM/Deep/Hash.pm 95.2 80.0 100.0 100.0 n/a 6.7 92.3
2689 Total 95.0 83.4 73.8 97.6 33.3 100.0 89.9
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.