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 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 }
110 if (scalar(@_) > 1) {
112 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
116 elsif ( ref $_[0] ) {
117 unless ( eval { local $SIG{'__DIE__'}; %{$_[0]} || 1 } ) {
118 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
123 $args = { file => shift };
131 # Class constructor method for Perl OO interface.
132 # Calls tie() and returns blessed reference to tied hash or array,
133 # providing a hybrid OO/tie interface.
136 my $args = $class->_get_args( @_ );
139 # Check if we want a tied hash or array.
142 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
143 $class = 'DBM::Deep::Array';
144 require DBM::Deep::Array;
145 tie @$self, $class, %$args;
148 $class = 'DBM::Deep::Hash';
149 require DBM::Deep::Hash;
150 tie %$self, $class, %$args;
153 return bless $self, $class;
158 # Setup $self and bless into this class.
163 # These are the defaults to be optionally overridden below
166 base_offset => length(SIG_FILE),
169 foreach my $param ( keys %$self ) {
170 next unless exists $args->{$param};
171 $self->{$param} = delete $args->{$param}
174 # locking implicitly enables autoflush
175 if ($args->{locking}) { $args->{autoflush} = 1; }
177 $self->{root} = exists $args->{root}
179 : DBM::Deep::_::Root->new( $args );
181 if (!defined($self->_fh)) { $self->_open(); }
188 require DBM::Deep::Hash;
189 return DBM::Deep::Hash->TIEHASH( @_ );
194 require DBM::Deep::Array;
195 return DBM::Deep::Array->TIEARRAY( @_ );
198 #XXX Unneeded now ...
204 # Open a fh to the database, create if nonexistent.
205 # Make sure file signature matches DBM::Deep spec.
207 my $self = $_[0]->_get_self;
209 if (defined($self->_fh)) { $self->_close(); }
211 my $flags = O_RDWR | O_CREAT | O_BINARY;
214 sysopen( $fh, $self->_root->{file}, $flags )
215 or $self->_throw_error( "Cannot sysopen file: " . $self->_root->{file} . ": $!" );
217 $self->_root->{fh} = $fh;
219 if ($self->_root->{autoflush}) {
220 my $old = select $fh;
225 seek($fh, 0 + $self->_root->{file_offset}, SEEK_SET);
228 my $bytes_read = read( $fh, $signature, length(SIG_FILE));
231 # File is empty -- write signature and master index
234 seek($fh, 0 + $self->_root->{file_offset}, SEEK_SET);
235 print( $fh SIG_FILE);
236 $self->_create_tag($self->_base_offset, $self->_type, chr(0) x $INDEX_SIZE);
238 my $plain_key = "[base]";
239 print( $fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
241 # Flush the filehandle
242 my $old_fh = select $fh;
243 my $old_af = $|; $| = 1; $| = $old_af;
246 my @stats = stat($fh);
247 $self->_root->{inode} = $stats[1];
248 $self->_root->{end} = $stats[7];
254 # Check signature was valid
256 unless ($signature eq SIG_FILE) {
258 return $self->_throw_error("Signature not found -- file is not a Deep DB");
261 my @stats = stat($fh);
262 $self->_root->{inode} = $stats[1];
263 $self->_root->{end} = $stats[7];
266 # Get our type from master index signature
268 my $tag = $self->_load_tag($self->_base_offset);
270 #XXX We probably also want to store the hash algorithm name and not assume anything
271 #XXX The cool thing would be to allow a different hashing algorithm at every level
274 return $self->_throw_error("Corrupted file, no master index record");
276 if ($self->{type} ne $tag->{signature}) {
277 return $self->_throw_error("File type mismatch");
287 my $self = $_[0]->_get_self;
288 close $self->_root->{fh} if $self->_root->{fh};
289 $self->_root->{fh} = undef;
294 # Given offset, signature and content, create tag and write to disk
296 my ($self, $offset, $sig, $content) = @_;
297 my $size = length($content);
301 seek($fh, $offset + $self->_root->{file_offset}, SEEK_SET);
302 print( $fh $sig . pack($DATA_LENGTH_PACK, $size) . $content );
304 if ($offset == $self->_root->{end}) {
305 $self->_root->{end} += SIG_SIZE + $DATA_LENGTH_SIZE + $size;
311 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
318 # Given offset, load single tag and return signature, size and data
325 seek($fh, $offset + $self->_root->{file_offset}, SEEK_SET);
326 if (eof $fh) { return undef; }
329 read( $fh, $b, SIG_SIZE + $DATA_LENGTH_SIZE );
330 my ($sig, $size) = unpack( "A $DATA_LENGTH_PACK", $b );
333 read( $fh, $buffer, $size);
338 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
345 # Given index tag, lookup single entry in index and return .
348 my ($tag, $index) = @_;
350 my $location = unpack($LONG_PACK, substr($tag->{content}, $index * $LONG_SIZE, $LONG_SIZE) );
351 if (!$location) { return; }
353 return $self->_load_tag( $location );
358 # Adds one key/value pair to bucket list, given offset, MD5 digest of key,
359 # plain (undigested) key and value.
362 my ($tag, $md5, $plain_key, $value) = @_;
363 my $keys = $tag->{content};
367 my $root = $self->_root;
369 my $is_dbm_deep = eval { local $SIG{'__DIE__'}; $value->isa( 'DBM::Deep' ) };
370 my $internal_ref = $is_dbm_deep && ($value->_root eq $root);
375 # Iterate through buckets, seeing if this is a new entry or a replace.
377 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
378 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
381 # Found empty bucket (end of list). Populate and exit loop.
385 $location = $internal_ref
386 ? $value->_base_offset
389 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $root->{file_offset}, SEEK_SET);
390 print( $fh $md5 . pack($LONG_PACK, $location) );
394 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
397 # Found existing bucket with same key. Replace with new value.
402 $location = $value->_base_offset;
403 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $root->{file_offset}, SEEK_SET);
404 print( $fh $md5 . pack($LONG_PACK, $location) );
408 seek($fh, $subloc + SIG_SIZE + $root->{file_offset}, SEEK_SET);
410 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
413 # If value is a hash, array, or raw value with equal or less size, we can
414 # reuse the same content area of the database. Otherwise, we have to create
415 # a new content area at the EOF.
418 my $r = Scalar::Util::reftype( $value ) || '';
419 if ( $r eq 'HASH' || $r eq 'ARRAY' ) {
420 $actual_length = $INDEX_SIZE;
422 # if autobless is enabled, must also take into consideration
423 # the class name, as it is stored along with key/value.
424 if ( $root->{autobless} ) {
425 my $value_class = Scalar::Util::blessed($value);
426 if ( defined $value_class && !$value->isa('DBM::Deep') ) {
427 $actual_length += length($value_class);
431 else { $actual_length = length($value); }
433 if ($actual_length <= $size) {
437 $location = $root->{end};
438 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE + $root->{file_offset}, SEEK_SET);
439 print( $fh pack($LONG_PACK, $location) );
447 # If this is an internal reference, return now.
448 # No need to write value or plain key
455 # If bucket didn't fit into list, split into a new index level
458 seek($fh, $tag->{ref_loc} + $root->{file_offset}, SEEK_SET);
459 print( $fh pack($LONG_PACK, $root->{end}) );
461 my $index_tag = $self->_create_tag($root->{end}, SIG_INDEX, chr(0) x $INDEX_SIZE);
464 $keys .= $md5 . pack($LONG_PACK, 0);
466 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
467 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
469 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
470 my $num = ord(substr($key, $tag->{ch} + 1, 1));
472 if ($offsets[$num]) {
473 my $offset = $offsets[$num] + SIG_SIZE + $DATA_LENGTH_SIZE;
474 seek($fh, $offset + $root->{file_offset}, SEEK_SET);
476 read( $fh, $subkeys, $BUCKET_LIST_SIZE);
478 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
479 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
481 seek($fh, $offset + ($k * $BUCKET_SIZE) + $root->{file_offset}, SEEK_SET);
482 print( $fh $key . pack($LONG_PACK, $old_subloc || $root->{end}) );
488 $offsets[$num] = $root->{end};
489 seek($fh, $index_tag->{offset} + ($num * $LONG_SIZE) + $root->{file_offset}, SEEK_SET);
490 print( $fh pack($LONG_PACK, $root->{end}) );
492 my $blist_tag = $self->_create_tag($root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
494 seek($fh, $blist_tag->{offset} + $root->{file_offset}, SEEK_SET);
495 print( $fh $key . pack($LONG_PACK, $old_subloc || $root->{end}) );
500 $location ||= $root->{end};
501 } # re-index bucket list
504 # Seek to content area and store signature, value and plaintext key
508 seek($fh, $location + $root->{file_offset}, SEEK_SET);
511 # Write signature based on content type, set content length and write actual value.
513 my $r = Scalar::Util::reftype($value) || '';
515 print( $fh TYPE_HASH );
516 print( $fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
517 $content_length = $INDEX_SIZE;
519 elsif ($r eq 'ARRAY') {
520 print( $fh TYPE_ARRAY );
521 print( $fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
522 $content_length = $INDEX_SIZE;
524 elsif (!defined($value)) {
525 print( $fh SIG_NULL );
526 print( $fh pack($DATA_LENGTH_PACK, 0) );
530 print( $fh SIG_DATA );
531 print( $fh pack($DATA_LENGTH_PACK, length($value)) . $value );
532 $content_length = length($value);
536 # Plain key is stored AFTER value, as keys are typically fetched less often.
538 print( $fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
541 # If value is blessed, preserve class name
543 if ( $root->{autobless} ) {
544 my $value_class = Scalar::Util::blessed($value);
545 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
547 # Blessed ref -- will restore later
550 print( $fh pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
551 $content_length += 1;
552 $content_length += $DATA_LENGTH_SIZE + length($value_class);
556 $content_length += 1;
561 # If this is a new content area, advance EOF counter
563 if ($location == $root->{end}) {
564 $root->{end} += SIG_SIZE;
565 $root->{end} += $DATA_LENGTH_SIZE + $content_length;
566 $root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
570 # If content is a hash or array, create new child DBM::Deep object and
571 # pass each key or element to it.
575 tie %$value, 'DBM::Deep', {
577 base_offset => $location,
582 elsif ($r eq 'ARRAY') {
584 tie @$value, 'DBM::Deep', {
586 base_offset => $location,
595 return $self->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
598 sub _get_bucket_value {
600 # Fetch single value given tag and MD5 digested key.
603 my ($tag, $md5) = @_;
604 my $keys = $tag->{content};
609 # Iterate through buckets, looking for a key match
612 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
613 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
614 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
618 # Hit end of list, no match
623 if ( $md5 ne $key ) {
628 # Found match -- seek to offset and read signature
631 seek($fh, $subloc + $self->_root->{file_offset}, SEEK_SET);
632 read( $fh, $signature, SIG_SIZE);
635 # If value is a hash or array, return new DBM::Deep object with correct offset
637 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
638 my $obj = DBM::Deep->new(
640 base_offset => $subloc,
644 if ($self->_root->{autobless}) {
646 # Skip over value and plain key to see if object needs
649 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
652 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
653 if ($size) { seek($fh, $size, SEEK_CUR); }
656 read( $fh, $bless_bit, 1);
657 if (ord($bless_bit)) {
659 # Yes, object needs to be re-blessed
662 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
663 if ($size) { read( $fh, $class_name, $size); }
664 if ($class_name) { $obj = bless( $obj, $class_name ); }
672 # Otherwise return actual value
674 elsif ($signature eq SIG_DATA) {
677 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
678 if ($size) { read( $fh, $value, $size); }
683 # Key exists, but content is null
693 # Delete single key/value pair given tag and MD5 digested key.
696 my ($tag, $md5) = @_;
697 my $keys = $tag->{content};
702 # Iterate through buckets, looking for a key match
705 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
706 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
707 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
711 # Hit end of list, no match
716 if ( $md5 ne $key ) {
721 # Matched key -- delete bucket and return
723 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $self->_root->{file_offset}, SEEK_SET);
724 print( $fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
725 print( $fh chr(0) x $BUCKET_SIZE );
735 # Check existence of single key given tag and MD5 digested key.
738 my ($tag, $md5) = @_;
739 my $keys = $tag->{content};
742 # Iterate through buckets, looking for a key match
745 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
746 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
747 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
751 # Hit end of list, no match
756 if ( $md5 ne $key ) {
761 # Matched key -- return true
769 sub _find_bucket_list {
771 # Locate offset for bucket list, given digested key
777 # Locate offset for bucket list using digest index system
780 my $tag = $self->_load_tag($self->_base_offset);
781 if (!$tag) { return; }
783 while ($tag->{signature} ne SIG_BLIST) {
784 $tag = $self->_index_lookup($tag, ord(substr($md5, $ch, 1)));
785 if (!$tag) { return; }
792 sub _traverse_index {
794 # Scan index and recursively step into deeper levels, looking for next key.
796 my ($self, $offset, $ch, $force_return_next) = @_;
797 $force_return_next = undef unless $force_return_next;
799 my $tag = $self->_load_tag( $offset );
803 if ($tag->{signature} ne SIG_BLIST) {
804 my $content = $tag->{content};
806 if ($self->{return_next}) { $start = 0; }
807 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
809 for (my $index = $start; $index < 256; $index++) {
810 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
812 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
813 if (defined($result)) { return $result; }
817 $self->{return_next} = 1;
820 elsif ($tag->{signature} eq SIG_BLIST) {
821 my $keys = $tag->{content};
822 if ($force_return_next) { $self->{return_next} = 1; }
825 # Iterate through buckets, looking for a key match
827 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
828 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
829 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
833 # End of bucket list -- return to outer loop
835 $self->{return_next} = 1;
838 elsif ($key eq $self->{prev_md5}) {
840 # Located previous key -- return next one found
842 $self->{return_next} = 1;
845 elsif ($self->{return_next}) {
847 # Seek to bucket location and skip over signature
849 seek($fh, $subloc + SIG_SIZE + $self->_root->{file_offset}, SEEK_SET);
852 # Skip over value to get to plain key
855 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
856 if ($size) { seek($fh, $size, SEEK_CUR); }
859 # Read in plain key and return as scalar
862 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
863 if ($size) { read( $fh, $plain_key, $size); }
869 $self->{return_next} = 1;
870 } # tag is a bucket list
877 # Locate next key, given digested previous one
879 my $self = $_[0]->_get_self;
881 $self->{prev_md5} = $_[1] ? $_[1] : undef;
882 $self->{return_next} = 0;
885 # If the previous key was not specifed, start at the top and
886 # return the first one found.
888 if (!$self->{prev_md5}) {
889 $self->{prev_md5} = chr(0) x $HASH_SIZE;
890 $self->{return_next} = 1;
893 return $self->_traverse_index( $self->_base_offset, 0 );
898 # If db locking is set, flock() the db file. If called multiple
899 # times before unlock(), then the same number of unlocks() must
900 # be called before the lock is released.
902 my $self = $_[0]->_get_self;
904 $type = LOCK_EX unless defined $type;
906 if (!defined($self->_fh)) { return; }
908 if ($self->_root->{locking}) {
909 if (!$self->_root->{locked}) {
910 flock($self->_fh, $type);
912 # refresh end counter in case file has changed size
913 my @stats = stat($self->_root->{file});
914 $self->_root->{end} = $stats[7];
916 # double-check file inode, in case another process
917 # has optimize()d our file while we were waiting.
918 if ($stats[1] != $self->_root->{inode}) {
919 $self->_open(); # re-open
920 flock($self->_fh, $type); # re-lock
921 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
924 $self->_root->{locked}++;
934 # If db locking is set, unlock the db file. See note in lock()
935 # regarding calling lock() multiple times.
937 my $self = $_[0]->_get_self;
939 if (!defined($self->_fh)) { return; }
941 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
942 $self->_root->{locked}--;
943 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
952 my $self = shift->_get_self;
953 my ($spot, $value) = @_;
958 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
959 my $type = $value->_type;
960 ${$spot} = $type eq TYPE_HASH ? {} : [];
961 $value->_copy_node( ${$spot} );
964 my $r = Scalar::Util::reftype( $value );
965 my $c = Scalar::Util::blessed( $value );
966 if ( $r eq 'ARRAY' ) {
967 ${$spot} = [ @{$value} ];
970 ${$spot} = { %{$value} };
972 ${$spot} = bless ${$spot}, $c
981 # Copy single level of keys or elements to new DB handle.
982 # Recurse for nested structures
984 my $self = shift->_get_self;
987 if ($self->_type eq TYPE_HASH) {
988 my $key = $self->first_key();
990 my $value = $self->get($key);
991 $self->_copy_value( \$db_temp->{$key}, $value );
992 $key = $self->next_key($key);
996 my $length = $self->length();
997 for (my $index = 0; $index < $length; $index++) {
998 my $value = $self->get($index);
999 $self->_copy_value( \$db_temp->[$index], $value );
1008 # Recursively export into standard Perl hashes and arrays.
1010 my $self = $_[0]->_get_self;
1013 if ($self->_type eq TYPE_HASH) { $temp = {}; }
1014 elsif ($self->_type eq TYPE_ARRAY) { $temp = []; }
1017 $self->_copy_node( $temp );
1025 # Recursively import Perl hash/array structure
1027 #XXX This use of ref() seems to be ok
1028 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
1030 my $self = $_[0]->_get_self;
1033 #XXX This use of ref() seems to be ok
1034 if (!ref($struct)) {
1036 # struct is not a reference, so just import based on our type
1040 if ($self->_type eq TYPE_HASH) { $struct = {@_}; }
1041 elsif ($self->_type eq TYPE_ARRAY) { $struct = [@_]; }
1044 my $r = Scalar::Util::reftype($struct) || '';
1045 if ($r eq "HASH" && $self->_type eq TYPE_HASH) {
1046 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
1048 elsif ($r eq "ARRAY" && $self->_type eq TYPE_ARRAY) {
1049 $self->push( @$struct );
1052 return $self->_throw_error("Cannot import: type mismatch");
1060 # Rebuild entire database into new file, then move
1061 # it back on top of original.
1063 my $self = $_[0]->_get_self;
1065 #XXX Need to create a new test for this
1066 # if ($self->_root->{links} > 1) {
1067 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
1070 my $db_temp = DBM::Deep->new(
1071 file => $self->_root->{file} . '.tmp',
1072 type => $self->_type
1075 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
1079 $self->_copy_node( $db_temp );
1083 # Attempt to copy user, group and permissions over to new file
1085 my @stats = stat($self->_fh);
1086 my $perms = $stats[2] & 07777;
1087 my $uid = $stats[4];
1088 my $gid = $stats[5];
1089 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
1090 chmod( $perms, $self->_root->{file} . '.tmp' );
1092 # q.v. perlport for more information on this variable
1093 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
1095 # Potential race condition when optmizing on Win32 with locking.
1096 # The Windows filesystem requires that the filehandle be closed
1097 # before it is overwritten with rename(). This could be redone
1104 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
1105 unlink $self->_root->{file} . '.tmp';
1107 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
1119 # Make copy of object and return
1121 my $self = $_[0]->_get_self;
1123 return DBM::Deep->new(
1124 type => $self->_type,
1125 base_offset => $self->_base_offset,
1126 root => $self->_root
1131 my %is_legal_filter = map {
1134 store_key store_value
1135 fetch_key fetch_value
1140 # Setup filter function for storing or fetching the key or value
1142 my $self = $_[0]->_get_self;
1143 my $type = lc $_[1];
1144 my $func = $_[2] ? $_[2] : undef;
1146 if ( $is_legal_filter{$type} ) {
1147 $self->_root->{"filter_$type"} = $func;
1161 # Get access to the root structure
1163 my $self = $_[0]->_get_self;
1164 return $self->{root};
1169 # Get access to the raw fh
1171 #XXX It will be useful, though, when we split out HASH and ARRAY
1172 my $self = $_[0]->_get_self;
1173 return $self->_root->{fh};
1178 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
1180 my $self = $_[0]->_get_self;
1181 return $self->{type};
1186 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
1188 my $self = $_[0]->_get_self;
1189 return $self->{base_offset};
1194 # Get last error string, or undef if no error
1197 ? ( $_[0]->_get_self->{root}->{error} or undef )
1207 # Store error string in self
1209 my $error_text = $_[1];
1211 if ( Scalar::Util::blessed $_[0] ) {
1212 my $self = $_[0]->_get_self;
1213 $self->_root->{error} = $error_text;
1215 unless ($self->_root->{debug}) {
1216 die "DBM::Deep: $error_text\n";
1219 warn "DBM::Deep: $error_text\n";
1223 die "DBM::Deep: $error_text\n";
1231 my $self = $_[0]->_get_self;
1233 undef $self->_root->{error};
1236 sub _precalc_sizes {
1238 # Precalculate index, bucket and bucket list sizes
1241 #XXX I don't like this ...
1242 set_pack() unless defined $LONG_SIZE;
1244 $INDEX_SIZE = 256 * $LONG_SIZE;
1245 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
1246 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
1251 # Set pack/unpack modes (see file header for more)
1253 my ($long_s, $long_p, $data_s, $data_p) = @_;
1255 $LONG_SIZE = $long_s ? $long_s : 4;
1256 $LONG_PACK = $long_p ? $long_p : 'N';
1258 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
1259 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
1266 # Set key digest function (default is MD5)
1268 my ($digest_func, $hash_size) = @_;
1270 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
1271 $HASH_SIZE = $hash_size ? $hash_size : 16;
1278 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
1283 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
1287 # tie() methods (hashes and arrays)
1292 # Store single hash key/value or array element in database.
1294 my $self = $_[0]->_get_self;
1297 # User may be storing a hash, in which case we do not want it run
1298 # through the filtering system
1299 my $value = ($self->_root->{filter_store_value} && !ref($_[2]))
1300 ? $self->_root->{filter_store_value}->($_[2])
1303 my $md5 = $DIGEST_FUNC->($key);
1306 # Make sure file is open
1308 if (!defined($self->_fh) && !$self->_open()) {
1312 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
1313 $self->_throw_error( 'Cannot write to a readonly filehandle' );
1317 # Request exclusive lock for writing
1319 $self->lock( LOCK_EX );
1321 my $fh = $self->_fh;
1324 # Locate offset for bucket list using digest index system
1326 my $tag = $self->_load_tag($self->_base_offset);
1328 $tag = $self->_create_tag($self->_base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1332 while ($tag->{signature} ne SIG_BLIST) {
1333 my $num = ord(substr($md5, $ch, 1));
1335 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1336 my $new_tag = $self->_index_lookup($tag, $num);
1339 seek($fh, $ref_loc + $self->_root->{file_offset}, SEEK_SET);
1340 print( $fh pack($LONG_PACK, $self->_root->{end}) );
1342 $tag = $self->_create_tag($self->_root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1344 $tag->{ref_loc} = $ref_loc;
1352 $tag->{ref_loc} = $ref_loc;
1359 # Add key/value to bucket list
1361 my $result = $self->_add_bucket( $tag, $md5, $key, $value );
1370 # Fetch single value or element given plain key or array index
1372 my $self = shift->_get_self;
1376 # Make sure file is open
1378 if (!defined($self->_fh)) { $self->_open(); }
1380 my $md5 = $DIGEST_FUNC->($key);
1383 # Request shared lock for reading
1385 $self->lock( LOCK_SH );
1387 my $tag = $self->_find_bucket_list( $md5 );
1394 # Get value from bucket list
1396 my $result = $self->_get_bucket_value( $tag, $md5 );
1400 #XXX What is ref() checking here?
1401 #YYY Filters only apply on scalar values, so the ref check is making
1402 #YYY sure the fetched bucket is a scalar, not a child hash or array.
1403 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
1404 ? $self->_root->{filter_fetch_value}->($result)
1410 # Delete single key/value pair or element given plain key or array index
1412 my $self = $_[0]->_get_self;
1415 my $md5 = $DIGEST_FUNC->($key);
1418 # Make sure file is open
1420 if (!defined($self->_fh)) { $self->_open(); }
1423 # Request exclusive lock for writing
1425 $self->lock( LOCK_EX );
1427 my $tag = $self->_find_bucket_list( $md5 );
1436 my $value = $self->_get_bucket_value( $tag, $md5 );
1437 if ($value && !ref($value) && $self->_root->{filter_fetch_value}) {
1438 $value = $self->_root->{filter_fetch_value}->($value);
1441 my $result = $self->_delete_bucket( $tag, $md5 );
1444 # If this object is an array and the key deleted was on the end of the stack,
1445 # decrement the length variable.
1455 # Check if a single key or element exists given plain key or array index
1457 my $self = $_[0]->_get_self;
1460 my $md5 = $DIGEST_FUNC->($key);
1463 # Make sure file is open
1465 if (!defined($self->_fh)) { $self->_open(); }
1468 # Request shared lock for reading
1470 $self->lock( LOCK_SH );
1472 my $tag = $self->_find_bucket_list( $md5 );
1475 # For some reason, the built-in exists() function returns '' for false
1483 # Check if bucket exists and return 1 or ''
1485 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1494 # Clear all keys from hash, or all elements from array.
1496 my $self = $_[0]->_get_self;
1499 # Make sure file is open
1501 if (!defined($self->_fh)) { $self->_open(); }
1504 # Request exclusive lock for writing
1506 $self->lock( LOCK_EX );
1508 my $fh = $self->_fh;
1510 seek($fh, $self->_base_offset + $self->_root->{file_offset}, SEEK_SET);
1516 $self->_create_tag($self->_base_offset, $self->_type, chr(0) x $INDEX_SIZE);
1524 # Public method aliases
1526 sub put { (shift)->STORE( @_ ) }
1527 sub store { (shift)->STORE( @_ ) }
1528 sub get { (shift)->FETCH( @_ ) }
1529 sub fetch { (shift)->FETCH( @_ ) }
1530 sub delete { (shift)->DELETE( @_ ) }
1531 sub exists { (shift)->EXISTS( @_ ) }
1532 sub clear { (shift)->CLEAR( @_ ) }
1534 package DBM::Deep::_::Root;
1548 filter_store_key => undef,
1549 filter_store_value => undef,
1550 filter_fetch_key => undef,
1551 filter_fetch_value => undef,
1557 if ( $self->{fh} && !$self->{file_offset} ) {
1558 $self->{file_offset} = tell( $self->{fh} );
1566 return unless $self;
1568 close $self->{fh} if $self->{fh};
1579 DBM::Deep - A pure perl multi-level hash/array DBM
1584 my $db = DBM::Deep->new( "foo.db" );
1586 $db->{key} = 'value'; # tie() style
1589 $db->put('key' => 'value'); # OO style
1590 print $db->get('key');
1592 # true multi-level support
1593 $db->{my_complex} = [
1594 'hello', { perl => 'rules' },
1600 A unique flat-file database module, written in pure perl. True
1601 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1602 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1603 handle millions of keys and unlimited hash levels without significant
1604 slow-down. Written from the ground-up in pure perl -- this is NOT a
1605 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1606 Mac OS X and Windows.
1610 Hopefully you are using Perl's excellent CPAN module, which will download
1611 and install the module for you. If not, get the tarball, and run these
1623 Construction can be done OO-style (which is the recommended way), or using
1624 Perl's tie() function. Both are examined here.
1626 =head2 OO CONSTRUCTION
1628 The recommended way to construct a DBM::Deep object is to use the new()
1629 method, which gets you a blessed, tied hash or array reference.
1631 my $db = DBM::Deep->new( "foo.db" );
1633 This opens a new database handle, mapped to the file "foo.db". If this
1634 file does not exist, it will automatically be created. DB files are
1635 opened in "r+" (read/write) mode, and the type of object returned is a
1636 hash, unless otherwise specified (see L<OPTIONS> below).
1638 You can pass a number of options to the constructor to specify things like
1639 locking, autoflush, etc. This is done by passing an inline hash:
1641 my $db = DBM::Deep->new(
1647 Notice that the filename is now specified I<inside> the hash with
1648 the "file" parameter, as opposed to being the sole argument to the
1649 constructor. This is required if any options are specified.
1650 See L<OPTIONS> below for the complete list.
1654 You can also start with an array instead of a hash. For this, you must
1655 specify the C<type> parameter:
1657 my $db = DBM::Deep->new(
1659 type => DBM::Deep->TYPE_ARRAY
1662 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1663 a new DB file. If you create a DBM::Deep object with an existing file, the
1664 C<type> will be loaded from the file header, and an error will be thrown if
1665 the wrong type is passed in.
1667 =head2 TIE CONSTRUCTION
1669 Alternately, you can create a DBM::Deep handle by using Perl's built-in
1670 tie() function. The object returned from tie() can be used to call methods,
1671 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
1672 file (as expected with most tie'd objects).
1675 my $db = tie %hash, "DBM::Deep", "foo.db";
1678 my $db = tie @array, "DBM::Deep", "bar.db";
1680 As with the OO constructor, you can replace the DB filename parameter with
1681 a hash containing one or more options (see L<OPTIONS> just below for the
1684 tie %hash, "DBM::Deep", {
1692 There are a number of options that can be passed in when constructing your
1693 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1699 Filename of the DB file to link the handle to. You can pass a full absolute
1700 filesystem path, partial path, or a plain filename if the file is in the
1701 current working directory. This is a required parameter (though q.v. fh).
1705 If you want, you can pass in the fh instead of the file. This is most useful for doing
1708 my $db = DBM::Deep->new( { fh => \*DATA } );
1710 You are responsible for making sure that the fh has been opened appropriately for your
1711 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
1712 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
1713 needs to read from the fh.
1717 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
1718 not need to set this. However, it's there if you want it.
1720 If you pass in fh and do not set this, it will be set appropriately.
1724 This parameter specifies what type of object to create, a hash or array. Use
1725 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1726 This only takes effect when beginning a new file. This is an optional
1727 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1731 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1732 function to lock the database in exclusive mode for writes, and shared mode for
1733 reads. Pass any true value to enable. This affects the base DB handle I<and
1734 any child hashes or arrays> that use the same DB file. This is an optional
1735 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1739 Specifies whether autoflush is to be enabled on the underlying filehandle.
1740 This obviously slows down write operations, but is required if you may have
1741 multiple processes accessing the same DB file (also consider enable I<locking>).
1742 Pass any true value to enable. This is an optional parameter, and defaults to 0
1747 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1748 restore them when fetched. This is an B<experimental> feature, and does have
1749 side-effects. Basically, when hashes are re-blessed into their original
1750 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1751 able to call any DBM::Deep methods on them. You have been warned.
1752 This is an optional parameter, and defaults to 0 (disabled).
1756 See L<FILTERS> below.
1760 Setting I<debug> mode will make all errors non-fatal, dump them out to
1761 STDERR, and continue on. This is for debugging purposes only, and probably
1762 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1764 B<NOTE>: This parameter is considered deprecated and should not be used anymore.
1768 =head1 TIE INTERFACE
1770 With DBM::Deep you can access your databases using Perl's standard hash/array
1771 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
1772 treat them as such. DBM::Deep will intercept all reads/writes and direct them
1773 to the right place -- the DB file. This has nothing to do with the
1774 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
1775 using regular hashes and arrays, rather than calling functions like C<get()>
1776 and C<put()> (although those work too). It is entirely up to you how to want
1777 to access your databases.
1781 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1782 or even nested hashes (or arrays) using standard Perl syntax:
1784 my $db = DBM::Deep->new( "foo.db" );
1786 $db->{mykey} = "myvalue";
1788 $db->{myhash}->{subkey} = "subvalue";
1790 print $db->{myhash}->{subkey} . "\n";
1792 You can even step through hash keys using the normal Perl C<keys()> function:
1794 foreach my $key (keys %$db) {
1795 print "$key: " . $db->{$key} . "\n";
1798 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1799 pushes them onto an array, all before the loop even begins. If you have an
1800 extra large hash, this may exhaust Perl's memory. Instead, consider using
1801 Perl's C<each()> function, which pulls keys/values one at a time, using very
1804 while (my ($key, $value) = each %$db) {
1805 print "$key: $value\n";
1808 Please note that when using C<each()>, you should always pass a direct
1809 hash reference, not a lookup. Meaning, you should B<never> do this:
1812 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1814 This causes an infinite loop, because for each iteration, Perl is calling
1815 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1816 it effectively keeps returning the first key over and over again. Instead,
1817 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1821 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1822 reference. This includes inserting, removing and manipulating elements,
1823 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1824 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1825 or simply be a nested array reference inside a hash. Example:
1827 my $db = DBM::Deep->new(
1828 file => "foo-array.db",
1829 type => DBM::Deep->TYPE_ARRAY
1833 push @$db, "bar", "baz";
1834 unshift @$db, "bah";
1836 my $last_elem = pop @$db; # baz
1837 my $first_elem = shift @$db; # bah
1838 my $second_elem = $db->[1]; # bar
1840 my $num_elements = scalar @$db;
1844 In addition to the I<tie()> interface, you can also use a standard OO interface
1845 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1846 array) has its own methods, but both types share the following common methods:
1847 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1851 =item * new() / clone()
1853 These are the constructor and copy-functions.
1855 =item * put() / store()
1857 Stores a new hash key/value pair, or sets an array element value. Takes two
1858 arguments, the hash key or array index, and the new value. The value can be
1859 a scalar, hash ref or array ref. Returns true on success, false on failure.
1861 $db->put("foo", "bar"); # for hashes
1862 $db->put(1, "bar"); # for arrays
1864 =item * get() / fetch()
1866 Fetches the value of a hash key or array element. Takes one argument: the hash
1867 key or array index. Returns a scalar, hash ref or array ref, depending on the
1870 my $value = $db->get("foo"); # for hashes
1871 my $value = $db->get(1); # for arrays
1875 Checks if a hash key or array index exists. Takes one argument: the hash key
1876 or array index. Returns true if it exists, false if not.
1878 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1879 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1883 Deletes one hash key/value pair or array element. Takes one argument: the hash
1884 key or array index. Returns true on success, false if not found. For arrays,
1885 the remaining elements located after the deleted element are NOT moved over.
1886 The deleted element is essentially just undefined, which is exactly how Perl's
1887 internal arrays work. Please note that the space occupied by the deleted
1888 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1889 below for details and workarounds.
1891 $db->delete("foo"); # for hashes
1892 $db->delete(1); # for arrays
1896 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1897 value. Please note that the space occupied by the deleted keys/values or
1898 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1899 details and workarounds.
1901 $db->clear(); # hashes or arrays
1903 =item * lock() / unlock()
1909 Recover lost disk space.
1911 =item * import() / export()
1913 Data going in and out.
1915 =item * set_digest() / set_pack() / set_filter()
1917 q.v. adjusting the interal parameters.
1919 =item * error() / clear_error()
1921 Error handling methods. These are deprecated and will be removed in 1.00.
1927 For hashes, DBM::Deep supports all the common methods described above, and the
1928 following additional methods: C<first_key()> and C<next_key()>.
1934 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1935 fetched in an undefined order (which appears random). Takes no arguments,
1936 returns the key as a scalar value.
1938 my $key = $db->first_key();
1942 Returns the "next" key in the hash, given the previous one as the sole argument.
1943 Returns undef if there are no more keys to be fetched.
1945 $key = $db->next_key($key);
1949 Here are some examples of using hashes:
1951 my $db = DBM::Deep->new( "foo.db" );
1953 $db->put("foo", "bar");
1954 print "foo: " . $db->get("foo") . "\n";
1956 $db->put("baz", {}); # new child hash ref
1957 $db->get("baz")->put("buz", "biz");
1958 print "buz: " . $db->get("baz")->get("buz") . "\n";
1960 my $key = $db->first_key();
1962 print "$key: " . $db->get($key) . "\n";
1963 $key = $db->next_key($key);
1966 if ($db->exists("foo")) { $db->delete("foo"); }
1970 For arrays, DBM::Deep supports all the common methods described above, and the
1971 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1972 C<unshift()> and C<splice()>.
1978 Returns the number of elements in the array. Takes no arguments.
1980 my $len = $db->length();
1984 Adds one or more elements onto the end of the array. Accepts scalars, hash
1985 refs or array refs. No return value.
1987 $db->push("foo", "bar", {});
1991 Fetches the last element in the array, and deletes it. Takes no arguments.
1992 Returns undef if array is empty. Returns the element value.
1994 my $elem = $db->pop();
1998 Fetches the first element in the array, deletes it, then shifts all the
1999 remaining elements over to take up the space. Returns the element value. This
2000 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
2003 my $elem = $db->shift();
2007 Inserts one or more elements onto the beginning of the array, shifting all
2008 existing elements over to make room. Accepts scalars, hash refs or array refs.
2009 No return value. This method is not recommended with large arrays -- see
2010 <LARGE ARRAYS> below for details.
2012 $db->unshift("foo", "bar", {});
2016 Performs exactly like Perl's built-in function of the same name. See L<perldoc
2017 -f splice> for usage -- it is too complicated to document here. This method is
2018 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
2022 Here are some examples of using arrays:
2024 my $db = DBM::Deep->new(
2026 type => DBM::Deep->TYPE_ARRAY
2029 $db->push("bar", "baz");
2030 $db->unshift("foo");
2033 my $len = $db->length();
2034 print "length: $len\n"; # 4
2036 for (my $k=0; $k<$len; $k++) {
2037 print "$k: " . $db->get($k) . "\n";
2040 $db->splice(1, 2, "biz", "baf");
2042 while (my $elem = shift @$db) {
2043 print "shifted: $elem\n";
2048 Enable automatic file locking by passing a true value to the C<locking>
2049 parameter when constructing your DBM::Deep object (see L<SETUP> above).
2051 my $db = DBM::Deep->new(
2056 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
2057 mode for writes, and shared mode for reads. This is required if you have
2058 multiple processes accessing the same database file, to avoid file corruption.
2059 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
2060 NFS> below for more.
2062 =head2 EXPLICIT LOCKING
2064 You can explicitly lock a database, so it remains locked for multiple
2065 transactions. This is done by calling the C<lock()> method, and passing an
2066 optional lock mode argument (defaults to exclusive mode). This is particularly
2067 useful for things like counters, where the current value needs to be fetched,
2068 then incremented, then stored again.
2071 my $counter = $db->get("counter");
2073 $db->put("counter", $counter);
2082 You can pass C<lock()> an optional argument, which specifies which mode to use
2083 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2084 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2085 same as the constants defined in Perl's C<Fcntl> module.
2087 $db->lock( DBM::Deep->LOCK_SH );
2091 =head1 IMPORTING/EXPORTING
2093 You can import existing complex structures by calling the C<import()> method,
2094 and export an entire database into an in-memory structure using the C<export()>
2095 method. Both are examined here.
2099 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2100 walking the structure and adding keys/elements to the database as you go,
2101 simply pass a reference to the C<import()> method. This recursively adds
2102 everything to an existing DBM::Deep object for you. Here is an example:
2107 array1 => [ "elem0", "elem1", "elem2" ],
2109 subkey1 => "subvalue1",
2110 subkey2 => "subvalue2"
2114 my $db = DBM::Deep->new( "foo.db" );
2115 $db->import( $struct );
2117 print $db->{key1} . "\n"; # prints "value1"
2119 This recursively imports the entire C<$struct> object into C<$db>, including
2120 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2121 keys are merged with the existing ones, replacing if they already exist.
2122 The C<import()> method can be called on any database level (not just the base
2123 level), and works with both hash and array DB types.
2125 B<Note:> Make sure your existing structure has no circular references in it.
2126 These will cause an infinite loop when importing.
2130 Calling the C<export()> method on an existing DBM::Deep object will return
2131 a reference to a new in-memory copy of the database. The export is done
2132 recursively, so all nested hashes/arrays are all exported to standard Perl
2133 objects. Here is an example:
2135 my $db = DBM::Deep->new( "foo.db" );
2137 $db->{key1} = "value1";
2138 $db->{key2} = "value2";
2140 $db->{hash1}->{subkey1} = "subvalue1";
2141 $db->{hash1}->{subkey2} = "subvalue2";
2143 my $struct = $db->export();
2145 print $struct->{key1} . "\n"; # prints "value1"
2147 This makes a complete copy of the database in memory, and returns a reference
2148 to it. The C<export()> method can be called on any database level (not just
2149 the base level), and works with both hash and array DB types. Be careful of
2150 large databases -- you can store a lot more data in a DBM::Deep object than an
2151 in-memory Perl structure.
2153 B<Note:> Make sure your database has no circular references in it.
2154 These will cause an infinite loop when exporting.
2158 DBM::Deep has a number of hooks where you can specify your own Perl function
2159 to perform filtering on incoming or outgoing data. This is a perfect
2160 way to extend the engine, and implement things like real-time compression or
2161 encryption. Filtering applies to the base DB level, and all child hashes /
2162 arrays. Filter hooks can be specified when your DBM::Deep object is first
2163 constructed, or by calling the C<set_filter()> method at any time. There are
2164 four available filter hooks, described below:
2168 =item * filter_store_key
2170 This filter is called whenever a hash key is stored. It
2171 is passed the incoming key, and expected to return a transformed key.
2173 =item * filter_store_value
2175 This filter is called whenever a hash key or array element is stored. It
2176 is passed the incoming value, and expected to return a transformed value.
2178 =item * filter_fetch_key
2180 This filter is called whenever a hash key is fetched (i.e. via
2181 C<first_key()> or C<next_key()>). It is passed the transformed key,
2182 and expected to return the plain key.
2184 =item * filter_fetch_value
2186 This filter is called whenever a hash key or array element is fetched.
2187 It is passed the transformed value, and expected to return the plain value.
2191 Here are the two ways to setup a filter hook:
2193 my $db = DBM::Deep->new(
2195 filter_store_value => \&my_filter_store,
2196 filter_fetch_value => \&my_filter_fetch
2201 $db->set_filter( "filter_store_value", \&my_filter_store );
2202 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2204 Your filter function will be called only when dealing with SCALAR keys or
2205 values. When nested hashes and arrays are being stored/fetched, filtering
2206 is bypassed. Filters are called as static functions, passed a single SCALAR
2207 argument, and expected to return a single SCALAR value. If you want to
2208 remove a filter, set the function reference to C<undef>:
2210 $db->set_filter( "filter_store_value", undef );
2212 =head2 REAL-TIME ENCRYPTION EXAMPLE
2214 Here is a working example that uses the I<Crypt::Blowfish> module to
2215 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2216 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2217 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2220 use Crypt::Blowfish;
2223 my $cipher = Crypt::CBC->new({
2224 'key' => 'my secret key',
2225 'cipher' => 'Blowfish',
2227 'regenerate_key' => 0,
2228 'padding' => 'space',
2232 my $db = DBM::Deep->new(
2233 file => "foo-encrypt.db",
2234 filter_store_key => \&my_encrypt,
2235 filter_store_value => \&my_encrypt,
2236 filter_fetch_key => \&my_decrypt,
2237 filter_fetch_value => \&my_decrypt,
2240 $db->{key1} = "value1";
2241 $db->{key2} = "value2";
2242 print "key1: " . $db->{key1} . "\n";
2243 print "key2: " . $db->{key2} . "\n";
2249 return $cipher->encrypt( $_[0] );
2252 return $cipher->decrypt( $_[0] );
2255 =head2 REAL-TIME COMPRESSION EXAMPLE
2257 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2258 compression / decompression of keys & values with DBM::Deep Filters.
2259 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2260 more on I<Compress::Zlib>.
2265 my $db = DBM::Deep->new(
2266 file => "foo-compress.db",
2267 filter_store_key => \&my_compress,
2268 filter_store_value => \&my_compress,
2269 filter_fetch_key => \&my_decompress,
2270 filter_fetch_value => \&my_decompress,
2273 $db->{key1} = "value1";
2274 $db->{key2} = "value2";
2275 print "key1: " . $db->{key1} . "\n";
2276 print "key2: " . $db->{key2} . "\n";
2282 return Compress::Zlib::memGzip( $_[0] ) ;
2285 return Compress::Zlib::memGunzip( $_[0] ) ;
2288 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2289 actually numerical index numbers, and are not filtered.
2291 =head1 ERROR HANDLING
2293 Most DBM::Deep methods return a true value for success, and call die() on
2294 failure. You can wrap calls in an eval block to catch the die. Also, the
2295 actual error message is stored in an internal scalar, which can be fetched by
2296 calling the C<error()> method.
2298 my $db = DBM::Deep->new( "foo.db" ); # create hash
2299 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2301 print $@; # prints error message
2302 print $db->error(); # prints error message
2304 You can then call C<clear_error()> to clear the current error state.
2308 If you set the C<debug> option to true when creating your DBM::Deep object,
2309 all errors are considered NON-FATAL, and dumped to STDERR. This should only
2310 be used for debugging purposes and not production work. DBM::Deep expects errors
2311 to be thrown, not propagated back up the stack.
2313 B<NOTE>: error() and clear_error() are considered deprecated and I<will> be removed
2314 in 1.00. Please don't use them. Instead, wrap all your functions with in eval-blocks.
2316 =head1 LARGEFILE SUPPORT
2318 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2319 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2320 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2321 by calling the static C<set_pack()> method before you do anything else.
2323 DBM::Deep::set_pack(8, 'Q');
2325 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2326 instead of 32-bit longs. After setting these values your DB files have a
2327 theoretical maximum size of 16 XB (exabytes).
2329 B<Note:> Changing these values will B<NOT> work for existing database files.
2330 Only change this for new files, and make sure it stays set consistently
2331 throughout the file's life. If you do set these values, you can no longer
2332 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2333 back to 32-bit mode.
2335 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2336 only a 32-bit Perl. However, I have received user reports that this does
2339 =head1 LOW-LEVEL ACCESS
2341 If you require low-level access to the underlying filehandle that DBM::Deep uses,
2342 you can call the C<_fh()> method, which returns the handle:
2344 my $fh = $db->_fh();
2346 This method can be called on the root level of the datbase, or any child
2347 hashes or arrays. All levels share a I<root> structure, which contains things
2348 like the filehandle, a reference counter, and all the options specified
2349 when you created the object. You can get access to this root structure by
2350 calling the C<root()> method.
2352 my $root = $db->_root();
2354 This is useful for changing options after the object has already been created,
2355 such as enabling/disabling locking, or debug modes. You can also
2356 store your own temporary user data in this structure (be wary of name
2357 collision), which is then accessible from any child hash or array.
2359 =head1 CUSTOM DIGEST ALGORITHM
2361 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2362 keys. However you can override this, and use another algorithm (such as SHA-256)
2363 or even write your own. But please note that DBM::Deep currently expects zero
2364 collisions, so your algorithm has to be I<perfect>, so to speak.
2365 Collision detection may be introduced in a later version.
2369 You can specify a custom digest algorithm by calling the static C<set_digest()>
2370 function, passing a reference to a subroutine, and the length of the algorithm's
2371 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2372 objects. Here is a working example that uses a 256-bit hash from the
2373 I<Digest::SHA256> module. Please see
2374 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2379 my $context = Digest::SHA256::new(256);
2381 DBM::Deep::set_digest( \&my_digest, 32 );
2383 my $db = DBM::Deep->new( "foo-sha.db" );
2385 $db->{key1} = "value1";
2386 $db->{key2} = "value2";
2387 print "key1: " . $db->{key1} . "\n";
2388 print "key2: " . $db->{key2} . "\n";
2394 return substr( $context->hash($_[0]), 0, 32 );
2397 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2398 of bytes you specify in the C<set_digest()> function (in this case 32).
2400 =head1 CIRCULAR REFERENCES
2402 DBM::Deep has B<experimental> support for circular references. Meaning you
2403 can have a nested hash key or array element that points to a parent object.
2404 This relationship is stored in the DB file, and is preserved between sessions.
2407 my $db = DBM::Deep->new( "foo.db" );
2410 $db->{circle} = $db; # ref to self
2412 print $db->{foo} . "\n"; # prints "foo"
2413 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2415 One catch is, passing the object to a function that recursively walks the
2416 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2417 C<export()> methods) will result in an infinite loop. The other catch is,
2418 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2419 or C<next_key()> methods), you will get the I<target object's key>, not the
2420 ref's key. This gets even more interesting with the above example, where
2421 the I<circle> key points to the base DB object, which technically doesn't
2422 have a key. So I made DBM::Deep return "[base]" as the key name in that
2425 =head1 CAVEATS / ISSUES / BUGS
2427 This section describes all the known issues with DBM::Deep. It you have found
2428 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2430 =head2 UNUSED SPACE RECOVERY
2432 One major caveat with DBM::Deep is that space occupied by existing keys and
2433 values is not recovered when they are deleted. Meaning if you keep deleting
2434 and adding new keys, your file will continuously grow. I am working on this,
2435 but in the meantime you can call the built-in C<optimize()> method from time to
2436 time (perhaps in a crontab or something) to recover all your unused space.
2438 $db->optimize(); # returns true on success
2440 This rebuilds the ENTIRE database into a new file, then moves it on top of
2441 the original. The new file will have no unused space, thus it will take up as
2442 little disk space as possible. Please note that this operation can take
2443 a long time for large files, and you need enough disk space to temporarily hold
2444 2 copies of your DB file. The temporary file is created in the same directory
2445 as the original, named with a ".tmp" extension, and is deleted when the
2446 operation completes. Oh, and if locking is enabled, the DB is automatically
2447 locked for the entire duration of the copy.
2449 B<WARNING:> Only call optimize() on the top-level node of the database, and
2450 make sure there are no child references lying around. DBM::Deep keeps a reference
2451 counter, and if it is greater than 1, optimize() will abort and return undef.
2453 =head2 FILE CORRUPTION
2455 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2456 for a 32-bit signature when opened, but other corruption in files can cause
2457 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2458 stuck in an infinite loop depending on the level of corruption. File write
2459 operations are not checked for failure (for speed), so if you happen to run
2460 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2461 be addressed in a later version of DBM::Deep.
2465 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2466 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2467 about setting up your NFS server with a locking daemon, then using lockf() to
2468 lock your files, but your mileage may vary there as well. From what I
2469 understand, there is no real way to do it. However, if you need access to the
2470 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
2471 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2473 =head2 COPYING OBJECTS
2475 Beware of copying tied objects in Perl. Very strange things can happen.
2476 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2477 returns a new, blessed, tied hash or array to the same level in the DB.
2479 my $copy = $db->clone();
2481 B<Note>: Since clone() here is cloning the object, not the database location, any
2482 modifications to either $db or $copy will be visible in both.
2486 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2487 These functions cause every element in the array to move, which can be murder
2488 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2489 a different location. This will be addressed in the forthcoming version 1.00.
2491 =head2 WRITEONLY FILES
2493 If you pass in a filehandle to new(), you may have opened it in either a readonly or
2494 writeonly mode. STORE will verify that the filehandle is writable. However, there
2495 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
2496 filehandle isn't readable, it's not clear what will happen. So, don't do that.
2500 This section discusses DBM::Deep's speed and memory usage.
2504 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2505 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2506 multi-level hash/array support, and cross-platform FTPable files. Even so,
2507 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2508 with huge databases. Here is some test data:
2510 Adding 1,000,000 keys to new DB file...
2512 At 100 keys, avg. speed is 2,703 keys/sec
2513 At 200 keys, avg. speed is 2,642 keys/sec
2514 At 300 keys, avg. speed is 2,598 keys/sec
2515 At 400 keys, avg. speed is 2,578 keys/sec
2516 At 500 keys, avg. speed is 2,722 keys/sec
2517 At 600 keys, avg. speed is 2,628 keys/sec
2518 At 700 keys, avg. speed is 2,700 keys/sec
2519 At 800 keys, avg. speed is 2,607 keys/sec
2520 At 900 keys, avg. speed is 2,190 keys/sec
2521 At 1,000 keys, avg. speed is 2,570 keys/sec
2522 At 2,000 keys, avg. speed is 2,417 keys/sec
2523 At 3,000 keys, avg. speed is 1,982 keys/sec
2524 At 4,000 keys, avg. speed is 1,568 keys/sec
2525 At 5,000 keys, avg. speed is 1,533 keys/sec
2526 At 6,000 keys, avg. speed is 1,787 keys/sec
2527 At 7,000 keys, avg. speed is 1,977 keys/sec
2528 At 8,000 keys, avg. speed is 2,028 keys/sec
2529 At 9,000 keys, avg. speed is 2,077 keys/sec
2530 At 10,000 keys, avg. speed is 2,031 keys/sec
2531 At 20,000 keys, avg. speed is 1,970 keys/sec
2532 At 30,000 keys, avg. speed is 2,050 keys/sec
2533 At 40,000 keys, avg. speed is 2,073 keys/sec
2534 At 50,000 keys, avg. speed is 1,973 keys/sec
2535 At 60,000 keys, avg. speed is 1,914 keys/sec
2536 At 70,000 keys, avg. speed is 2,091 keys/sec
2537 At 80,000 keys, avg. speed is 2,103 keys/sec
2538 At 90,000 keys, avg. speed is 1,886 keys/sec
2539 At 100,000 keys, avg. speed is 1,970 keys/sec
2540 At 200,000 keys, avg. speed is 2,053 keys/sec
2541 At 300,000 keys, avg. speed is 1,697 keys/sec
2542 At 400,000 keys, avg. speed is 1,838 keys/sec
2543 At 500,000 keys, avg. speed is 1,941 keys/sec
2544 At 600,000 keys, avg. speed is 1,930 keys/sec
2545 At 700,000 keys, avg. speed is 1,735 keys/sec
2546 At 800,000 keys, avg. speed is 1,795 keys/sec
2547 At 900,000 keys, avg. speed is 1,221 keys/sec
2548 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2550 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2551 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2552 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2553 Run time was 12 min 3 sec.
2557 One of the great things about DBM::Deep is that it uses very little memory.
2558 Even with huge databases (1,000,000+ keys) you will not see much increased
2559 memory on your process. DBM::Deep relies solely on the filesystem for storing
2560 and fetching data. Here is output from I</usr/bin/top> before even opening a
2563 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2564 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2566 Basically the process is taking 2,716K of memory. And here is the same
2567 process after storing and fetching 1,000,000 keys:
2569 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2570 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2572 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2573 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2575 =head1 DB FILE FORMAT
2577 In case you were interested in the underlying DB file format, it is documented
2578 here in this section. You don't need to know this to use the module, it's just
2579 included for reference.
2583 DBM::Deep files always start with a 32-bit signature to identify the file type.
2584 This is at offset 0. The signature is "DPDB" in network byte order. This is
2585 checked for when the file is opened and an error will be thrown if it's not found.
2589 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2590 has a standard header containing the type of data, the length of data, and then
2591 the data itself. The type is a single character (1 byte), the length is a
2592 32-bit unsigned long in network byte order, and the data is, well, the data.
2593 Here is how it unfolds:
2597 Immediately after the 32-bit file signature is the I<Master Index> record.
2598 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2599 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2600 depending on how the DBM::Deep object was constructed.
2602 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2603 number). The first 8-bit char of the MD5 signature is the offset into the
2604 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2605 index element is a file offset of the next tag for the key/element in question,
2606 which is usually a I<Bucket List> tag (see below).
2608 The next tag I<could> be another index, depending on how many keys/elements
2609 exist. See L<RE-INDEXING> below for details.
2613 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2614 file offsets to where the actual data is stored. It starts with a standard
2615 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2616 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2617 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2618 When the list fills up, a I<Re-Index> operation is performed (See
2619 L<RE-INDEXING> below).
2623 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2624 index/value pair (in array mode). It starts with a standard tag header with
2625 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2626 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2627 header. The size reported in the tag header is only for the value, but then,
2628 just after the value is another size (32-bit unsigned long) and then the plain
2629 key itself. Since the value is likely to be fetched more often than the plain
2630 key, I figured it would be I<slightly> faster to store the value first.
2632 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2633 record for the nested structure, where the process begins all over again.
2637 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2638 exhausted. Then, when another key/element comes in, the list is converted to a
2639 new index record. However, this index will look at the next char in the MD5
2640 hash, and arrange new Bucket List pointers accordingly. This process is called
2641 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2642 17 (16 + new one) keys/elements are removed from the old Bucket List and
2643 inserted into the new index. Several new Bucket Lists are created in the
2644 process, as a new MD5 char from the key is being examined (it is unlikely that
2645 the keys will all share the same next char of their MD5s).
2647 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2648 when the Bucket Lists will turn into indexes, but the first round tends to
2649 happen right around 4,000 keys. You will see a I<slight> decrease in
2650 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2651 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2652 right around 900,000 keys. This process can continue nearly indefinitely --
2653 right up until the point the I<MD5> signatures start colliding with each other,
2654 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2655 getting struck by lightning while you are walking to cash in your tickets.
2656 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2657 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2658 this is 340 unodecillion, but don't quote me).
2662 When a new key/element is stored, the key (or index number) is first run through
2663 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2664 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2665 for the first char of the signature (in this case I<b0>). If it does not exist,
2666 a new I<Bucket List> is created for our key (and the next 15 future keys that
2667 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2668 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2669 this point, unless we are replacing an existing I<Bucket>), where the actual
2670 data will be stored.
2674 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2675 (or index number), then walking along the indexes. If there are enough
2676 keys/elements in this DB level, there might be nested indexes, each linked to
2677 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2678 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2679 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2680 plain key are stored.
2682 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2683 methods. In this process the indexes are walked systematically, and each key
2684 fetched in increasing MD5 order (which is why it appears random). Once the
2685 I<Bucket> is found, the value is skipped and the plain key returned instead.
2686 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2687 alphabetically sorted. This only happens on an index-level -- as soon as the
2688 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2689 so it's pretty much undefined how the keys will come out -- just like Perl's
2692 =head1 CODE COVERAGE
2694 We use B<Devel::Cover> to test the code coverage of our tests, below is the
2695 B<Devel::Cover> report on this module's test suite.
2697 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2698 File stmt bran cond sub pod time total
2699 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2700 blib/lib/DBM/Deep.pm 95.2 83.8 70.0 98.2 100.0 58.0 91.0
2701 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 n/a 26.7 98.0
2702 blib/lib/DBM/Deep/Hash.pm 95.3 80.0 100.0 100.0 n/a 15.3 92.4
2703 Total 96.2 84.8 74.4 98.8 100.0 100.0 92.4
2704 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2706 =head1 MORE INFORMATION
2708 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
2709 or send email to L<DBM-Deep@googlegroups.com>.
2713 Joseph Huckaby, L<jhuckaby@cpan.org>
2715 Rob Kinyon, L<rkinyon@cpan.org>
2717 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2721 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2722 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2726 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
2727 This is free software, you may use it and distribute it under the
2728 same terms as Perl itself.