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_NULL () { 'N' }
94 sub SIG_DATA () { 'D' }
95 sub SIG_INDEX () { 'I' }
96 sub SIG_BLIST () { 'B' }
100 # Setup constants for users to pass to new()
102 sub TYPE_HASH () { SIG_HASH }
103 sub TYPE_ARRAY () { SIG_ARRAY }
109 if (scalar(@_) > 1) {
111 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
115 elsif ( ref $_[0] ) {
116 unless ( eval { local $SIG{'__DIE__'}; %{$_[0]} || 1 } ) {
117 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
122 $args = { file => shift };
130 # Class constructor method for Perl OO interface.
131 # Calls tie() and returns blessed reference to tied hash or array,
132 # providing a hybrid OO/tie interface.
135 my $args = $class->_get_args( @_ );
138 # Check if we want a tied hash or array.
141 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
142 $class = 'DBM::Deep::Array';
143 require DBM::Deep::Array;
144 tie @$self, $class, %$args;
147 $class = 'DBM::Deep::Hash';
148 require DBM::Deep::Hash;
149 tie %$self, $class, %$args;
152 return bless $self, $class;
157 # Setup $self and bless into this class.
162 # These are the defaults to be optionally overridden below
165 base_offset => length(SIG_FILE),
168 foreach my $param ( keys %$self ) {
169 next unless exists $args->{$param};
170 $self->{$param} = delete $args->{$param}
173 # locking implicitly enables autoflush
174 if ($args->{locking}) { $args->{autoflush} = 1; }
176 $self->{root} = exists $args->{root}
178 : DBM::Deep::_::Root->new( $args );
180 if (!defined($self->_fh)) { $self->_open(); }
187 require DBM::Deep::Hash;
188 return DBM::Deep::Hash->TIEHASH( @_ );
193 require DBM::Deep::Array;
194 return DBM::Deep::Array->TIEARRAY( @_ );
197 #XXX Unneeded now ...
203 # Open a fh to the database, create if nonexistent.
204 # Make sure file signature matches DBM::Deep spec.
206 my $self = $_[0]->_get_self;
210 if (defined($self->_fh)) { $self->_close(); }
212 my $flags = O_RDWR | O_CREAT | O_BINARY;
215 sysopen( $fh, $self->_root->{file}, $flags )
216 or $self->_throw_error( "Cannot sysopen file: " . $self->_root->{file} . ": $!" );
218 $self->_root->{fh} = $fh;
220 if ($self->_root->{autoflush}) {
221 my $old = select $fh;
226 seek($fh, 0 + $self->_root->{file_offset}, SEEK_SET);
229 my $bytes_read = read( $fh, $signature, length(SIG_FILE));
232 # File is empty -- write signature and master index
235 seek($fh, 0 + $self->_root->{file_offset}, SEEK_SET);
236 print( $fh SIG_FILE);
237 $self->_create_tag($self->_base_offset, $self->_type, chr(0) x $INDEX_SIZE);
239 my $plain_key = "[base]";
240 print( $fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
242 # Flush the filehandle
243 my $old_fh = select $fh;
244 my $old_af = $|; $| = 1; $| = $old_af;
247 my @stats = stat($fh);
248 $self->_root->{inode} = $stats[1];
249 $self->_root->{end} = $stats[7];
255 # Check signature was valid
257 unless ($signature eq SIG_FILE) {
259 return $self->_throw_error("Signature not found -- file is not a Deep DB");
262 my @stats = stat($fh);
263 $self->_root->{inode} = $stats[1];
264 $self->_root->{end} = $stats[7];
267 # Get our type from master index signature
269 my $tag = $self->_load_tag($self->_base_offset);
271 #XXX We probably also want to store the hash algorithm name and not assume anything
272 #XXX The cool thing would be to allow a different hashing algorithm at every level
275 return $self->_throw_error("Corrupted file, no master index record");
277 if ($self->{type} ne $tag->{signature}) {
278 return $self->_throw_error("File type mismatch");
288 my $self = $_[0]->_get_self;
289 close $self->_root->{fh} if $self->_root->{fh};
290 $self->_root->{fh} = undef;
295 # Given offset, signature and content, create tag and write to disk
297 my ($self, $offset, $sig, $content) = @_;
298 my $size = length($content);
304 seek($fh, $offset + $self->_root->{file_offset}, SEEK_SET);
305 print( $fh $sig . pack($DATA_LENGTH_PACK, $size) . $content );
307 if ($offset == $self->_root->{end}) {
308 $self->_root->{end} += SIG_SIZE + $DATA_LENGTH_SIZE + $size;
314 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
321 # Given offset, load single tag and return signature, size and data
330 seek($fh, $offset + $self->_root->{file_offset}, SEEK_SET);
331 if (eof $fh) { return undef; }
334 read( $fh, $b, SIG_SIZE + $DATA_LENGTH_SIZE );
335 my ($sig, $size) = unpack( "A $DATA_LENGTH_PACK", $b );
338 read( $fh, $buffer, $size);
343 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
350 # Given index tag, lookup single entry in index and return .
353 my ($tag, $index) = @_;
355 my $location = unpack($LONG_PACK, substr($tag->{content}, $index * $LONG_SIZE, $LONG_SIZE) );
356 if (!$location) { return; }
358 return $self->_load_tag( $location );
363 # Adds one key/value pair to bucket list, given offset, MD5 digest of key,
364 # plain (undigested) key and value.
367 my ($tag, $md5, $plain_key, $value) = @_;
368 my $keys = $tag->{content};
374 # This verifies that only supported values will be stored.
376 my $r = Scalar::Util::reftype( $value );
379 last if $r eq 'HASH';
380 last if $r eq 'ARRAY';
383 "Storage of variables of type '$r' is not supported."
387 my $root = $self->_root;
389 my $is_dbm_deep = eval { local $SIG{'__DIE__'}; $value->isa( 'DBM::Deep' ) };
390 my $internal_ref = $is_dbm_deep && ($value->_root eq $root);
395 # Iterate through buckets, seeing if this is a new entry or a replace.
397 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
398 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
401 # Found empty bucket (end of list). Populate and exit loop.
405 $location = $internal_ref
406 ? $value->_base_offset
409 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $root->{file_offset}, SEEK_SET);
410 print( $fh $md5 . pack($LONG_PACK, $location) );
414 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
417 # Found existing bucket with same key. Replace with new value.
422 $location = $value->_base_offset;
423 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $root->{file_offset}, SEEK_SET);
424 print( $fh $md5 . pack($LONG_PACK, $location) );
428 seek($fh, $subloc + SIG_SIZE + $root->{file_offset}, SEEK_SET);
430 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
433 # If value is a hash, array, or raw value with equal or less size, we can
434 # reuse the same content area of the database. Otherwise, we have to create
435 # a new content area at the EOF.
438 my $r = Scalar::Util::reftype( $value ) || '';
439 if ( $r eq 'HASH' || $r eq 'ARRAY' ) {
440 $actual_length = $INDEX_SIZE;
442 # if autobless is enabled, must also take into consideration
443 # the class name, as it is stored along with key/value.
444 if ( $root->{autobless} ) {
445 my $value_class = Scalar::Util::blessed($value);
446 if ( defined $value_class && !$value->isa('DBM::Deep') ) {
447 $actual_length += length($value_class);
451 else { $actual_length = length($value); }
453 if ($actual_length <= ($size || 0)) {
457 $location = $root->{end};
458 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE + $root->{file_offset}, SEEK_SET);
459 print( $fh pack($LONG_PACK, $location) );
467 # If this is an internal reference, return now.
468 # No need to write value or plain key
475 # If bucket didn't fit into list, split into a new index level
478 seek($fh, $tag->{ref_loc} + $root->{file_offset}, SEEK_SET);
479 print( $fh pack($LONG_PACK, $root->{end}) );
481 my $index_tag = $self->_create_tag($root->{end}, SIG_INDEX, chr(0) x $INDEX_SIZE);
484 $keys .= $md5 . pack($LONG_PACK, 0);
486 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
487 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
489 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
490 my $num = ord(substr($key, $tag->{ch} + 1, 1));
492 if ($offsets[$num]) {
493 my $offset = $offsets[$num] + SIG_SIZE + $DATA_LENGTH_SIZE;
494 seek($fh, $offset + $root->{file_offset}, SEEK_SET);
496 read( $fh, $subkeys, $BUCKET_LIST_SIZE);
498 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
499 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
501 seek($fh, $offset + ($k * $BUCKET_SIZE) + $root->{file_offset}, SEEK_SET);
502 print( $fh $key . pack($LONG_PACK, $old_subloc || $root->{end}) );
508 $offsets[$num] = $root->{end};
509 seek($fh, $index_tag->{offset} + ($num * $LONG_SIZE) + $root->{file_offset}, SEEK_SET);
510 print( $fh pack($LONG_PACK, $root->{end}) );
512 my $blist_tag = $self->_create_tag($root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
514 seek($fh, $blist_tag->{offset} + $root->{file_offset}, SEEK_SET);
515 print( $fh $key . pack($LONG_PACK, $old_subloc || $root->{end}) );
520 $location ||= $root->{end};
521 } # re-index bucket list
524 # Seek to content area and store signature, value and plaintext key
528 seek($fh, $location + $root->{file_offset}, SEEK_SET);
531 # Write signature based on content type, set content length and write actual value.
533 my $r = Scalar::Util::reftype($value) || '';
535 if ( !$internal_ref && tied %{$value} ) {
536 return $self->_throw_error("Cannot store a tied value");
538 print( $fh TYPE_HASH );
539 print( $fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
540 $content_length = $INDEX_SIZE;
542 elsif ($r eq 'ARRAY') {
543 if ( !$internal_ref && tied @{$value} ) {
544 return $self->_throw_error("Cannot store a tied value");
546 print( $fh TYPE_ARRAY );
547 print( $fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
548 $content_length = $INDEX_SIZE;
550 elsif (!defined($value)) {
551 print( $fh SIG_NULL );
552 print( $fh pack($DATA_LENGTH_PACK, 0) );
556 print( $fh SIG_DATA );
557 print( $fh pack($DATA_LENGTH_PACK, length($value)) . $value );
558 $content_length = length($value);
562 # Plain key is stored AFTER value, as keys are typically fetched less often.
564 print( $fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
567 # If value is blessed, preserve class name
569 if ( $root->{autobless} ) {
570 my $value_class = Scalar::Util::blessed($value);
571 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
573 # Blessed ref -- will restore later
576 print( $fh pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
577 $content_length += 1;
578 $content_length += $DATA_LENGTH_SIZE + length($value_class);
582 $content_length += 1;
587 # If this is a new content area, advance EOF counter
589 if ($location == $root->{end}) {
590 $root->{end} += SIG_SIZE;
591 $root->{end} += $DATA_LENGTH_SIZE + $content_length;
592 $root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
596 # If content is a hash or array, create new child DBM::Deep object and
597 # pass each key or element to it.
601 tie %$value, 'DBM::Deep', {
603 base_offset => $location,
608 elsif ($r eq 'ARRAY') {
610 tie @$value, 'DBM::Deep', {
612 base_offset => $location,
621 return $self->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
624 sub _get_bucket_value {
626 # Fetch single value given tag and MD5 digested key.
629 my ($tag, $md5) = @_;
630 my $keys = $tag->{content};
637 # Iterate through buckets, looking for a key match
640 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
641 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
642 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
646 # Hit end of list, no match
651 if ( $md5 ne $key ) {
656 # Found match -- seek to offset and read signature
659 seek($fh, $subloc + $self->_root->{file_offset}, SEEK_SET);
660 read( $fh, $signature, SIG_SIZE);
663 # If value is a hash or array, return new DBM::Deep object with correct offset
665 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
666 my $obj = DBM::Deep->new(
668 base_offset => $subloc,
672 if ($self->_root->{autobless}) {
674 # Skip over value and plain key to see if object needs
677 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
680 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
681 if ($size) { seek($fh, $size, SEEK_CUR); }
684 read( $fh, $bless_bit, 1);
685 if (ord($bless_bit)) {
687 # Yes, object needs to be re-blessed
690 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
691 if ($size) { read( $fh, $class_name, $size); }
692 if ($class_name) { $obj = bless( $obj, $class_name ); }
700 # Otherwise return actual value
702 elsif ($signature eq SIG_DATA) {
705 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
706 if ($size) { read( $fh, $value, $size); }
711 # Key exists, but content is null
721 # Delete single key/value pair given tag and MD5 digested key.
724 my ($tag, $md5) = @_;
725 my $keys = $tag->{content};
732 # Iterate through buckets, looking for a key match
735 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
736 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
737 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
741 # Hit end of list, no match
746 if ( $md5 ne $key ) {
751 # Matched key -- delete bucket and return
753 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $self->_root->{file_offset}, SEEK_SET);
754 print( $fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
755 print( $fh chr(0) x $BUCKET_SIZE );
765 # Check existence of single key given tag and MD5 digested key.
768 my ($tag, $md5) = @_;
769 my $keys = $tag->{content};
772 # Iterate through buckets, looking for a key match
775 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
776 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
777 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
781 # Hit end of list, no match
786 if ( $md5 ne $key ) {
791 # Matched key -- return true
799 sub _find_bucket_list {
801 # Locate offset for bucket list, given digested key
807 # Locate offset for bucket list using digest index system
810 my $tag = $self->_load_tag($self->_base_offset);
811 if (!$tag) { return; }
813 while ($tag->{signature} ne SIG_BLIST) {
814 $tag = $self->_index_lookup($tag, ord(substr($md5, $ch, 1)));
815 if (!$tag) { return; }
822 sub _traverse_index {
824 # Scan index and recursively step into deeper levels, looking for next key.
826 my ($self, $offset, $ch, $force_return_next) = @_;
827 $force_return_next = undef unless $force_return_next;
831 my $tag = $self->_load_tag( $offset );
835 if ($tag->{signature} ne SIG_BLIST) {
836 my $content = $tag->{content};
838 if ($self->{return_next}) { $start = 0; }
839 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
841 for (my $index = $start; $index < 256; $index++) {
842 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
844 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
845 if (defined($result)) { return $result; }
849 $self->{return_next} = 1;
852 elsif ($tag->{signature} eq SIG_BLIST) {
853 my $keys = $tag->{content};
854 if ($force_return_next) { $self->{return_next} = 1; }
857 # Iterate through buckets, looking for a key match
859 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
860 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
861 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
865 # End of bucket list -- return to outer loop
867 $self->{return_next} = 1;
870 elsif ($key eq $self->{prev_md5}) {
872 # Located previous key -- return next one found
874 $self->{return_next} = 1;
877 elsif ($self->{return_next}) {
879 # Seek to bucket location and skip over signature
881 seek($fh, $subloc + SIG_SIZE + $self->_root->{file_offset}, SEEK_SET);
884 # Skip over value to get to plain key
887 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
888 if ($size) { seek($fh, $size, SEEK_CUR); }
891 # Read in plain key and return as scalar
894 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
895 if ($size) { read( $fh, $plain_key, $size); }
901 $self->{return_next} = 1;
902 } # tag is a bucket list
909 # Locate next key, given digested previous one
911 my $self = $_[0]->_get_self;
913 $self->{prev_md5} = $_[1] ? $_[1] : undef;
914 $self->{return_next} = 0;
917 # If the previous key was not specifed, start at the top and
918 # return the first one found.
920 if (!$self->{prev_md5}) {
921 $self->{prev_md5} = chr(0) x $HASH_SIZE;
922 $self->{return_next} = 1;
925 return $self->_traverse_index( $self->_base_offset, 0 );
930 # If db locking is set, flock() the db file. If called multiple
931 # times before unlock(), then the same number of unlocks() must
932 # be called before the lock is released.
934 my $self = $_[0]->_get_self;
936 $type = LOCK_EX unless defined $type;
938 if (!defined($self->_fh)) { return; }
940 if ($self->_root->{locking}) {
941 if (!$self->_root->{locked}) {
942 flock($self->_fh, $type);
944 # refresh end counter in case file has changed size
945 my @stats = stat($self->_root->{file});
946 $self->_root->{end} = $stats[7];
948 # double-check file inode, in case another process
949 # has optimize()d our file while we were waiting.
950 if ($stats[1] != $self->_root->{inode}) {
951 $self->_open(); # re-open
952 flock($self->_fh, $type); # re-lock
953 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
956 $self->_root->{locked}++;
966 # If db locking is set, unlock the db file. See note in lock()
967 # regarding calling lock() multiple times.
969 my $self = $_[0]->_get_self;
971 if (!defined($self->_fh)) { return; }
973 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
974 $self->_root->{locked}--;
975 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
984 my $self = shift->_get_self;
985 my ($spot, $value) = @_;
990 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
991 my $type = $value->_type;
992 ${$spot} = $type eq TYPE_HASH ? {} : [];
993 $value->_copy_node( ${$spot} );
996 my $r = Scalar::Util::reftype( $value );
997 my $c = Scalar::Util::blessed( $value );
998 if ( $r eq 'ARRAY' ) {
999 ${$spot} = [ @{$value} ];
1002 ${$spot} = { %{$value} };
1004 ${$spot} = bless ${$spot}, $c
1013 # Copy single level of keys or elements to new DB handle.
1014 # Recurse for nested structures
1016 my $self = shift->_get_self;
1019 if ($self->_type eq TYPE_HASH) {
1020 my $key = $self->first_key();
1022 my $value = $self->get($key);
1023 $self->_copy_value( \$db_temp->{$key}, $value );
1024 $key = $self->next_key($key);
1028 my $length = $self->length();
1029 for (my $index = 0; $index < $length; $index++) {
1030 my $value = $self->get($index);
1031 $self->_copy_value( \$db_temp->[$index], $value );
1040 # Recursively export into standard Perl hashes and arrays.
1042 my $self = $_[0]->_get_self;
1045 if ($self->_type eq TYPE_HASH) { $temp = {}; }
1046 elsif ($self->_type eq TYPE_ARRAY) { $temp = []; }
1049 $self->_copy_node( $temp );
1057 # Recursively import Perl hash/array structure
1059 #XXX This use of ref() seems to be ok
1060 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
1062 my $self = $_[0]->_get_self;
1065 #XXX This use of ref() seems to be ok
1066 if (!ref($struct)) {
1068 # struct is not a reference, so just import based on our type
1072 if ($self->_type eq TYPE_HASH) { $struct = {@_}; }
1073 elsif ($self->_type eq TYPE_ARRAY) { $struct = [@_]; }
1076 my $r = Scalar::Util::reftype($struct) || '';
1077 if ($r eq "HASH" && $self->_type eq TYPE_HASH) {
1078 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
1080 elsif ($r eq "ARRAY" && $self->_type eq TYPE_ARRAY) {
1081 $self->push( @$struct );
1084 return $self->_throw_error("Cannot import: type mismatch");
1092 # Rebuild entire database into new file, then move
1093 # it back on top of original.
1095 my $self = $_[0]->_get_self;
1097 #XXX Need to create a new test for this
1098 # if ($self->_root->{links} > 1) {
1099 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
1102 my $db_temp = DBM::Deep->new(
1103 file => $self->_root->{file} . '.tmp',
1104 type => $self->_type
1107 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
1111 $self->_copy_node( $db_temp );
1115 # Attempt to copy user, group and permissions over to new file
1117 my @stats = stat($self->_fh);
1118 my $perms = $stats[2] & 07777;
1119 my $uid = $stats[4];
1120 my $gid = $stats[5];
1121 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
1122 chmod( $perms, $self->_root->{file} . '.tmp' );
1124 # q.v. perlport for more information on this variable
1125 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
1127 # Potential race condition when optmizing on Win32 with locking.
1128 # The Windows filesystem requires that the filehandle be closed
1129 # before it is overwritten with rename(). This could be redone
1136 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
1137 unlink $self->_root->{file} . '.tmp';
1139 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
1151 # Make copy of object and return
1153 my $self = $_[0]->_get_self;
1155 return DBM::Deep->new(
1156 type => $self->_type,
1157 base_offset => $self->_base_offset,
1158 root => $self->_root
1163 my %is_legal_filter = map {
1166 store_key store_value
1167 fetch_key fetch_value
1172 # Setup filter function for storing or fetching the key or value
1174 my $self = $_[0]->_get_self;
1175 my $type = lc $_[1];
1176 my $func = $_[2] ? $_[2] : undef;
1178 if ( $is_legal_filter{$type} ) {
1179 $self->_root->{"filter_$type"} = $func;
1193 # Get access to the root structure
1195 my $self = $_[0]->_get_self;
1196 return $self->{root};
1201 # Get access to the raw fh
1203 #XXX It will be useful, though, when we split out HASH and ARRAY
1204 my $self = $_[0]->_get_self;
1205 return $self->_root->{fh};
1210 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
1212 my $self = $_[0]->_get_self;
1213 return $self->{type};
1218 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
1220 my $self = $_[0]->_get_self;
1221 return $self->{base_offset};
1226 # Get last error string, or undef if no error
1229 ? ( $_[0]->_get_self->{root}->{error} or undef )
1239 # Store error string in self
1241 my $error_text = $_[1];
1243 if ( Scalar::Util::blessed $_[0] ) {
1244 my $self = $_[0]->_get_self;
1245 $self->_root->{error} = $error_text;
1247 unless ($self->_root->{debug}) {
1248 die "DBM::Deep: $error_text\n";
1251 warn "DBM::Deep: $error_text\n";
1255 die "DBM::Deep: $error_text\n";
1263 my $self = $_[0]->_get_self;
1265 undef $self->_root->{error};
1268 sub _precalc_sizes {
1270 # Precalculate index, bucket and bucket list sizes
1273 #XXX I don't like this ...
1274 set_pack() unless defined $LONG_SIZE;
1276 $INDEX_SIZE = 256 * $LONG_SIZE;
1277 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
1278 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
1283 # Set pack/unpack modes (see file header for more)
1285 my ($long_s, $long_p, $data_s, $data_p) = @_;
1287 $LONG_SIZE = $long_s ? $long_s : 4;
1288 $LONG_PACK = $long_p ? $long_p : 'N';
1290 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
1291 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
1298 # Set key digest function (default is MD5)
1300 my ($digest_func, $hash_size) = @_;
1302 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
1303 $HASH_SIZE = $hash_size ? $hash_size : 16;
1310 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
1315 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
1319 # tie() methods (hashes and arrays)
1324 # Store single hash key/value or array element in database.
1326 my $self = $_[0]->_get_self;
1331 # User may be storing a hash, in which case we do not want it run
1332 # through the filtering system
1333 my $value = ($self->_root->{filter_store_value} && !ref($_[2]))
1334 ? $self->_root->{filter_store_value}->($_[2])
1337 my $md5 = $DIGEST_FUNC->($key);
1340 # Make sure file is open
1342 if (!defined($self->_fh) && !$self->_open()) {
1346 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
1347 $self->_throw_error( 'Cannot write to a readonly filehandle' );
1351 # Request exclusive lock for writing
1353 $self->lock( LOCK_EX );
1355 my $fh = $self->_fh;
1358 # Locate offset for bucket list using digest index system
1360 my $tag = $self->_load_tag($self->_base_offset);
1362 $tag = $self->_create_tag($self->_base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1366 while ($tag->{signature} ne SIG_BLIST) {
1367 my $num = ord(substr($md5, $ch, 1));
1369 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1370 my $new_tag = $self->_index_lookup($tag, $num);
1373 seek($fh, $ref_loc + $self->_root->{file_offset}, SEEK_SET);
1374 print( $fh pack($LONG_PACK, $self->_root->{end}) );
1376 $tag = $self->_create_tag($self->_root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1378 $tag->{ref_loc} = $ref_loc;
1386 $tag->{ref_loc} = $ref_loc;
1393 # Add key/value to bucket list
1395 my $result = $self->_add_bucket( $tag, $md5, $key, $value );
1404 # Fetch single value or element given plain key or array index
1406 my $self = shift->_get_self;
1410 # Make sure file is open
1412 if (!defined($self->_fh)) { $self->_open(); }
1414 my $md5 = $DIGEST_FUNC->($key);
1417 # Request shared lock for reading
1419 $self->lock( LOCK_SH );
1421 my $tag = $self->_find_bucket_list( $md5 );
1428 # Get value from bucket list
1430 my $result = $self->_get_bucket_value( $tag, $md5 );
1434 #XXX What is ref() checking here?
1435 #YYY Filters only apply on scalar values, so the ref check is making
1436 #YYY sure the fetched bucket is a scalar, not a child hash or array.
1437 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
1438 ? $self->_root->{filter_fetch_value}->($result)
1444 # Delete single key/value pair or element given plain key or array index
1446 my $self = $_[0]->_get_self;
1449 my $md5 = $DIGEST_FUNC->($key);
1452 # Make sure file is open
1454 if (!defined($self->_fh)) { $self->_open(); }
1457 # Request exclusive lock for writing
1459 $self->lock( LOCK_EX );
1461 my $tag = $self->_find_bucket_list( $md5 );
1470 my $value = $self->_get_bucket_value( $tag, $md5 );
1471 if ($value && !ref($value) && $self->_root->{filter_fetch_value}) {
1472 $value = $self->_root->{filter_fetch_value}->($value);
1475 my $result = $self->_delete_bucket( $tag, $md5 );
1478 # If this object is an array and the key deleted was on the end of the stack,
1479 # decrement the length variable.
1489 # Check if a single key or element exists given plain key or array index
1491 my $self = $_[0]->_get_self;
1494 my $md5 = $DIGEST_FUNC->($key);
1497 # Make sure file is open
1499 if (!defined($self->_fh)) { $self->_open(); }
1502 # Request shared lock for reading
1504 $self->lock( LOCK_SH );
1506 my $tag = $self->_find_bucket_list( $md5 );
1509 # For some reason, the built-in exists() function returns '' for false
1517 # Check if bucket exists and return 1 or ''
1519 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1528 # Clear all keys from hash, or all elements from array.
1530 my $self = $_[0]->_get_self;
1533 # Make sure file is open
1535 if (!defined($self->_fh)) { $self->_open(); }
1538 # Request exclusive lock for writing
1540 $self->lock( LOCK_EX );
1542 my $fh = $self->_fh;
1544 seek($fh, $self->_base_offset + $self->_root->{file_offset}, SEEK_SET);
1550 $self->_create_tag($self->_base_offset, $self->_type, chr(0) x $INDEX_SIZE);
1558 # Public method aliases
1560 sub put { (shift)->STORE( @_ ) }
1561 sub store { (shift)->STORE( @_ ) }
1562 sub get { (shift)->FETCH( @_ ) }
1563 sub fetch { (shift)->FETCH( @_ ) }
1564 sub delete { (shift)->DELETE( @_ ) }
1565 sub exists { (shift)->EXISTS( @_ ) }
1566 sub clear { (shift)->CLEAR( @_ ) }
1568 package DBM::Deep::_::Root;
1582 filter_store_key => undef,
1583 filter_store_value => undef,
1584 filter_fetch_key => undef,
1585 filter_fetch_value => undef,
1591 if ( $self->{fh} && !$self->{file_offset} ) {
1592 $self->{file_offset} = tell( $self->{fh} );
1600 return unless $self;
1602 close $self->{fh} if $self->{fh};
1613 DBM::Deep - A pure perl multi-level hash/array DBM
1618 my $db = DBM::Deep->new( "foo.db" );
1620 $db->{key} = 'value'; # tie() style
1623 $db->put('key' => 'value'); # OO style
1624 print $db->get('key');
1626 # true multi-level support
1627 $db->{my_complex} = [
1628 'hello', { perl => 'rules' },
1634 A unique flat-file database module, written in pure perl. True
1635 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1636 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1637 handle millions of keys and unlimited hash levels without significant
1638 slow-down. Written from the ground-up in pure perl -- this is NOT a
1639 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1640 Mac OS X and Windows.
1644 Hopefully you are using Perl's excellent CPAN module, which will download
1645 and install the module for you. If not, get the tarball, and run these
1657 Construction can be done OO-style (which is the recommended way), or using
1658 Perl's tie() function. Both are examined here.
1660 =head2 OO CONSTRUCTION
1662 The recommended way to construct a DBM::Deep object is to use the new()
1663 method, which gets you a blessed, tied hash or array reference.
1665 my $db = DBM::Deep->new( "foo.db" );
1667 This opens a new database handle, mapped to the file "foo.db". If this
1668 file does not exist, it will automatically be created. DB files are
1669 opened in "r+" (read/write) mode, and the type of object returned is a
1670 hash, unless otherwise specified (see L<OPTIONS> below).
1672 You can pass a number of options to the constructor to specify things like
1673 locking, autoflush, etc. This is done by passing an inline hash:
1675 my $db = DBM::Deep->new(
1681 Notice that the filename is now specified I<inside> the hash with
1682 the "file" parameter, as opposed to being the sole argument to the
1683 constructor. This is required if any options are specified.
1684 See L<OPTIONS> below for the complete list.
1688 You can also start with an array instead of a hash. For this, you must
1689 specify the C<type> parameter:
1691 my $db = DBM::Deep->new(
1693 type => DBM::Deep->TYPE_ARRAY
1696 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1697 a new DB file. If you create a DBM::Deep object with an existing file, the
1698 C<type> will be loaded from the file header, and an error will be thrown if
1699 the wrong type is passed in.
1701 =head2 TIE CONSTRUCTION
1703 Alternately, you can create a DBM::Deep handle by using Perl's built-in
1704 tie() function. The object returned from tie() can be used to call methods,
1705 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
1706 file (as expected with most tie'd objects).
1709 my $db = tie %hash, "DBM::Deep", "foo.db";
1712 my $db = tie @array, "DBM::Deep", "bar.db";
1714 As with the OO constructor, you can replace the DB filename parameter with
1715 a hash containing one or more options (see L<OPTIONS> just below for the
1718 tie %hash, "DBM::Deep", {
1726 There are a number of options that can be passed in when constructing your
1727 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1733 Filename of the DB file to link the handle to. You can pass a full absolute
1734 filesystem path, partial path, or a plain filename if the file is in the
1735 current working directory. This is a required parameter (though q.v. fh).
1739 If you want, you can pass in the fh instead of the file. This is most useful for doing
1742 my $db = DBM::Deep->new( { fh => \*DATA } );
1744 You are responsible for making sure that the fh has been opened appropriately for your
1745 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
1746 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
1747 needs to read from the fh.
1751 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
1752 not need to set this. However, it's there if you want it.
1754 If you pass in fh and do not set this, it will be set appropriately.
1758 This parameter specifies what type of object to create, a hash or array. Use
1759 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1760 This only takes effect when beginning a new file. This is an optional
1761 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1765 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1766 function to lock the database in exclusive mode for writes, and shared mode for
1767 reads. Pass any true value to enable. This affects the base DB handle I<and
1768 any child hashes or arrays> that use the same DB file. This is an optional
1769 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1773 Specifies whether autoflush is to be enabled on the underlying filehandle.
1774 This obviously slows down write operations, but is required if you may have
1775 multiple processes accessing the same DB file (also consider enable I<locking>).
1776 Pass any true value to enable. This is an optional parameter, and defaults to 0
1781 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1782 restore them when fetched. This is an B<experimental> feature, and does have
1783 side-effects. Basically, when hashes are re-blessed into their original
1784 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1785 able to call any DBM::Deep methods on them. You have been warned.
1786 This is an optional parameter, and defaults to 0 (disabled).
1790 See L<FILTERS> below.
1794 Setting I<debug> mode will make all errors non-fatal, dump them out to
1795 STDERR, and continue on. This is for debugging purposes only, and probably
1796 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1798 B<NOTE>: This parameter is considered deprecated and should not be used anymore.
1802 =head1 TIE INTERFACE
1804 With DBM::Deep you can access your databases using Perl's standard hash/array
1805 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
1806 treat them as such. DBM::Deep will intercept all reads/writes and direct them
1807 to the right place -- the DB file. This has nothing to do with the
1808 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
1809 using regular hashes and arrays, rather than calling functions like C<get()>
1810 and C<put()> (although those work too). It is entirely up to you how to want
1811 to access your databases.
1815 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1816 or even nested hashes (or arrays) using standard Perl syntax:
1818 my $db = DBM::Deep->new( "foo.db" );
1820 $db->{mykey} = "myvalue";
1822 $db->{myhash}->{subkey} = "subvalue";
1824 print $db->{myhash}->{subkey} . "\n";
1826 You can even step through hash keys using the normal Perl C<keys()> function:
1828 foreach my $key (keys %$db) {
1829 print "$key: " . $db->{$key} . "\n";
1832 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1833 pushes them onto an array, all before the loop even begins. If you have an
1834 extra large hash, this may exhaust Perl's memory. Instead, consider using
1835 Perl's C<each()> function, which pulls keys/values one at a time, using very
1838 while (my ($key, $value) = each %$db) {
1839 print "$key: $value\n";
1842 Please note that when using C<each()>, you should always pass a direct
1843 hash reference, not a lookup. Meaning, you should B<never> do this:
1846 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1848 This causes an infinite loop, because for each iteration, Perl is calling
1849 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1850 it effectively keeps returning the first key over and over again. Instead,
1851 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1855 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1856 reference. This includes inserting, removing and manipulating elements,
1857 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1858 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1859 or simply be a nested array reference inside a hash. Example:
1861 my $db = DBM::Deep->new(
1862 file => "foo-array.db",
1863 type => DBM::Deep->TYPE_ARRAY
1867 push @$db, "bar", "baz";
1868 unshift @$db, "bah";
1870 my $last_elem = pop @$db; # baz
1871 my $first_elem = shift @$db; # bah
1872 my $second_elem = $db->[1]; # bar
1874 my $num_elements = scalar @$db;
1878 In addition to the I<tie()> interface, you can also use a standard OO interface
1879 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1880 array) has its own methods, but both types share the following common methods:
1881 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1885 =item * new() / clone()
1887 These are the constructor and copy-functions.
1889 =item * put() / store()
1891 Stores a new hash key/value pair, or sets an array element value. Takes two
1892 arguments, the hash key or array index, and the new value. The value can be
1893 a scalar, hash ref or array ref. Returns true on success, false on failure.
1895 $db->put("foo", "bar"); # for hashes
1896 $db->put(1, "bar"); # for arrays
1898 =item * get() / fetch()
1900 Fetches the value of a hash key or array element. Takes one argument: the hash
1901 key or array index. Returns a scalar, hash ref or array ref, depending on the
1904 my $value = $db->get("foo"); # for hashes
1905 my $value = $db->get(1); # for arrays
1909 Checks if a hash key or array index exists. Takes one argument: the hash key
1910 or array index. Returns true if it exists, false if not.
1912 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1913 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1917 Deletes one hash key/value pair or array element. Takes one argument: the hash
1918 key or array index. Returns true on success, false if not found. For arrays,
1919 the remaining elements located after the deleted element are NOT moved over.
1920 The deleted element is essentially just undefined, which is exactly how Perl's
1921 internal arrays work. Please note that the space occupied by the deleted
1922 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1923 below for details and workarounds.
1925 $db->delete("foo"); # for hashes
1926 $db->delete(1); # for arrays
1930 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1931 value. Please note that the space occupied by the deleted keys/values or
1932 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1933 details and workarounds.
1935 $db->clear(); # hashes or arrays
1937 =item * lock() / unlock()
1943 Recover lost disk space.
1945 =item * import() / export()
1947 Data going in and out.
1949 =item * set_digest() / set_pack() / set_filter()
1951 q.v. adjusting the interal parameters.
1953 =item * error() / clear_error()
1955 Error handling methods. These are deprecated and will be removed in 1.00.
1961 For hashes, DBM::Deep supports all the common methods described above, and the
1962 following additional methods: C<first_key()> and C<next_key()>.
1968 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1969 fetched in an undefined order (which appears random). Takes no arguments,
1970 returns the key as a scalar value.
1972 my $key = $db->first_key();
1976 Returns the "next" key in the hash, given the previous one as the sole argument.
1977 Returns undef if there are no more keys to be fetched.
1979 $key = $db->next_key($key);
1983 Here are some examples of using hashes:
1985 my $db = DBM::Deep->new( "foo.db" );
1987 $db->put("foo", "bar");
1988 print "foo: " . $db->get("foo") . "\n";
1990 $db->put("baz", {}); # new child hash ref
1991 $db->get("baz")->put("buz", "biz");
1992 print "buz: " . $db->get("baz")->get("buz") . "\n";
1994 my $key = $db->first_key();
1996 print "$key: " . $db->get($key) . "\n";
1997 $key = $db->next_key($key);
2000 if ($db->exists("foo")) { $db->delete("foo"); }
2004 For arrays, DBM::Deep supports all the common methods described above, and the
2005 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
2006 C<unshift()> and C<splice()>.
2012 Returns the number of elements in the array. Takes no arguments.
2014 my $len = $db->length();
2018 Adds one or more elements onto the end of the array. Accepts scalars, hash
2019 refs or array refs. No return value.
2021 $db->push("foo", "bar", {});
2025 Fetches the last element in the array, and deletes it. Takes no arguments.
2026 Returns undef if array is empty. Returns the element value.
2028 my $elem = $db->pop();
2032 Fetches the first element in the array, deletes it, then shifts all the
2033 remaining elements over to take up the space. Returns the element value. This
2034 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
2037 my $elem = $db->shift();
2041 Inserts one or more elements onto the beginning of the array, shifting all
2042 existing elements over to make room. Accepts scalars, hash refs or array refs.
2043 No return value. This method is not recommended with large arrays -- see
2044 <LARGE ARRAYS> below for details.
2046 $db->unshift("foo", "bar", {});
2050 Performs exactly like Perl's built-in function of the same name. See L<perldoc
2051 -f splice> for usage -- it is too complicated to document here. This method is
2052 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
2056 Here are some examples of using arrays:
2058 my $db = DBM::Deep->new(
2060 type => DBM::Deep->TYPE_ARRAY
2063 $db->push("bar", "baz");
2064 $db->unshift("foo");
2067 my $len = $db->length();
2068 print "length: $len\n"; # 4
2070 for (my $k=0; $k<$len; $k++) {
2071 print "$k: " . $db->get($k) . "\n";
2074 $db->splice(1, 2, "biz", "baf");
2076 while (my $elem = shift @$db) {
2077 print "shifted: $elem\n";
2082 Enable automatic file locking by passing a true value to the C<locking>
2083 parameter when constructing your DBM::Deep object (see L<SETUP> above).
2085 my $db = DBM::Deep->new(
2090 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
2091 mode for writes, and shared mode for reads. This is required if you have
2092 multiple processes accessing the same database file, to avoid file corruption.
2093 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
2094 NFS> below for more.
2096 =head2 EXPLICIT LOCKING
2098 You can explicitly lock a database, so it remains locked for multiple
2099 transactions. This is done by calling the C<lock()> method, and passing an
2100 optional lock mode argument (defaults to exclusive mode). This is particularly
2101 useful for things like counters, where the current value needs to be fetched,
2102 then incremented, then stored again.
2105 my $counter = $db->get("counter");
2107 $db->put("counter", $counter);
2116 You can pass C<lock()> an optional argument, which specifies which mode to use
2117 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2118 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2119 same as the constants defined in Perl's C<Fcntl> module.
2121 $db->lock( DBM::Deep->LOCK_SH );
2125 =head1 IMPORTING/EXPORTING
2127 You can import existing complex structures by calling the C<import()> method,
2128 and export an entire database into an in-memory structure using the C<export()>
2129 method. Both are examined here.
2133 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2134 walking the structure and adding keys/elements to the database as you go,
2135 simply pass a reference to the C<import()> method. This recursively adds
2136 everything to an existing DBM::Deep object for you. Here is an example:
2141 array1 => [ "elem0", "elem1", "elem2" ],
2143 subkey1 => "subvalue1",
2144 subkey2 => "subvalue2"
2148 my $db = DBM::Deep->new( "foo.db" );
2149 $db->import( $struct );
2151 print $db->{key1} . "\n"; # prints "value1"
2153 This recursively imports the entire C<$struct> object into C<$db>, including
2154 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2155 keys are merged with the existing ones, replacing if they already exist.
2156 The C<import()> method can be called on any database level (not just the base
2157 level), and works with both hash and array DB types.
2159 B<Note:> Make sure your existing structure has no circular references in it.
2160 These will cause an infinite loop when importing.
2164 Calling the C<export()> method on an existing DBM::Deep object will return
2165 a reference to a new in-memory copy of the database. The export is done
2166 recursively, so all nested hashes/arrays are all exported to standard Perl
2167 objects. Here is an example:
2169 my $db = DBM::Deep->new( "foo.db" );
2171 $db->{key1} = "value1";
2172 $db->{key2} = "value2";
2174 $db->{hash1}->{subkey1} = "subvalue1";
2175 $db->{hash1}->{subkey2} = "subvalue2";
2177 my $struct = $db->export();
2179 print $struct->{key1} . "\n"; # prints "value1"
2181 This makes a complete copy of the database in memory, and returns a reference
2182 to it. The C<export()> method can be called on any database level (not just
2183 the base level), and works with both hash and array DB types. Be careful of
2184 large databases -- you can store a lot more data in a DBM::Deep object than an
2185 in-memory Perl structure.
2187 B<Note:> Make sure your database has no circular references in it.
2188 These will cause an infinite loop when exporting.
2192 DBM::Deep has a number of hooks where you can specify your own Perl function
2193 to perform filtering on incoming or outgoing data. This is a perfect
2194 way to extend the engine, and implement things like real-time compression or
2195 encryption. Filtering applies to the base DB level, and all child hashes /
2196 arrays. Filter hooks can be specified when your DBM::Deep object is first
2197 constructed, or by calling the C<set_filter()> method at any time. There are
2198 four available filter hooks, described below:
2202 =item * filter_store_key
2204 This filter is called whenever a hash key is stored. It
2205 is passed the incoming key, and expected to return a transformed key.
2207 =item * filter_store_value
2209 This filter is called whenever a hash key or array element is stored. It
2210 is passed the incoming value, and expected to return a transformed value.
2212 =item * filter_fetch_key
2214 This filter is called whenever a hash key is fetched (i.e. via
2215 C<first_key()> or C<next_key()>). It is passed the transformed key,
2216 and expected to return the plain key.
2218 =item * filter_fetch_value
2220 This filter is called whenever a hash key or array element is fetched.
2221 It is passed the transformed value, and expected to return the plain value.
2225 Here are the two ways to setup a filter hook:
2227 my $db = DBM::Deep->new(
2229 filter_store_value => \&my_filter_store,
2230 filter_fetch_value => \&my_filter_fetch
2235 $db->set_filter( "filter_store_value", \&my_filter_store );
2236 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2238 Your filter function will be called only when dealing with SCALAR keys or
2239 values. When nested hashes and arrays are being stored/fetched, filtering
2240 is bypassed. Filters are called as static functions, passed a single SCALAR
2241 argument, and expected to return a single SCALAR value. If you want to
2242 remove a filter, set the function reference to C<undef>:
2244 $db->set_filter( "filter_store_value", undef );
2246 =head2 REAL-TIME ENCRYPTION EXAMPLE
2248 Here is a working example that uses the I<Crypt::Blowfish> module to
2249 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2250 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2251 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2254 use Crypt::Blowfish;
2257 my $cipher = Crypt::CBC->new({
2258 'key' => 'my secret key',
2259 'cipher' => 'Blowfish',
2261 'regenerate_key' => 0,
2262 'padding' => 'space',
2266 my $db = DBM::Deep->new(
2267 file => "foo-encrypt.db",
2268 filter_store_key => \&my_encrypt,
2269 filter_store_value => \&my_encrypt,
2270 filter_fetch_key => \&my_decrypt,
2271 filter_fetch_value => \&my_decrypt,
2274 $db->{key1} = "value1";
2275 $db->{key2} = "value2";
2276 print "key1: " . $db->{key1} . "\n";
2277 print "key2: " . $db->{key2} . "\n";
2283 return $cipher->encrypt( $_[0] );
2286 return $cipher->decrypt( $_[0] );
2289 =head2 REAL-TIME COMPRESSION EXAMPLE
2291 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2292 compression / decompression of keys & values with DBM::Deep Filters.
2293 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2294 more on I<Compress::Zlib>.
2299 my $db = DBM::Deep->new(
2300 file => "foo-compress.db",
2301 filter_store_key => \&my_compress,
2302 filter_store_value => \&my_compress,
2303 filter_fetch_key => \&my_decompress,
2304 filter_fetch_value => \&my_decompress,
2307 $db->{key1} = "value1";
2308 $db->{key2} = "value2";
2309 print "key1: " . $db->{key1} . "\n";
2310 print "key2: " . $db->{key2} . "\n";
2316 return Compress::Zlib::memGzip( $_[0] ) ;
2319 return Compress::Zlib::memGunzip( $_[0] ) ;
2322 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2323 actually numerical index numbers, and are not filtered.
2325 =head1 ERROR HANDLING
2327 Most DBM::Deep methods return a true value for success, and call die() on
2328 failure. You can wrap calls in an eval block to catch the die. Also, the
2329 actual error message is stored in an internal scalar, which can be fetched by
2330 calling the C<error()> method.
2332 my $db = DBM::Deep->new( "foo.db" ); # create hash
2333 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2335 print $@; # prints error message
2336 print $db->error(); # prints error message
2338 You can then call C<clear_error()> to clear the current error state.
2342 If you set the C<debug> option to true when creating your DBM::Deep object,
2343 all errors are considered NON-FATAL, and dumped to STDERR. This should only
2344 be used for debugging purposes and not production work. DBM::Deep expects errors
2345 to be thrown, not propagated back up the stack.
2347 B<NOTE>: error() and clear_error() are considered deprecated and I<will> be removed
2348 in 1.00. Please don't use them. Instead, wrap all your functions with in eval-blocks.
2350 =head1 LARGEFILE SUPPORT
2352 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2353 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2354 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2355 by calling the static C<set_pack()> method before you do anything else.
2357 DBM::Deep::set_pack(8, 'Q');
2359 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2360 instead of 32-bit longs. After setting these values your DB files have a
2361 theoretical maximum size of 16 XB (exabytes).
2363 B<Note:> Changing these values will B<NOT> work for existing database files.
2364 Only change this for new files, and make sure it stays set consistently
2365 throughout the file's life. If you do set these values, you can no longer
2366 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2367 back to 32-bit mode.
2369 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2370 only a 32-bit Perl. However, I have received user reports that this does
2373 =head1 LOW-LEVEL ACCESS
2375 If you require low-level access to the underlying filehandle that DBM::Deep uses,
2376 you can call the C<_fh()> method, which returns the handle:
2378 my $fh = $db->_fh();
2380 This method can be called on the root level of the datbase, or any child
2381 hashes or arrays. All levels share a I<root> structure, which contains things
2382 like the filehandle, a reference counter, and all the options specified
2383 when you created the object. You can get access to this root structure by
2384 calling the C<root()> method.
2386 my $root = $db->_root();
2388 This is useful for changing options after the object has already been created,
2389 such as enabling/disabling locking, or debug modes. You can also
2390 store your own temporary user data in this structure (be wary of name
2391 collision), which is then accessible from any child hash or array.
2393 =head1 CUSTOM DIGEST ALGORITHM
2395 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2396 keys. However you can override this, and use another algorithm (such as SHA-256)
2397 or even write your own. But please note that DBM::Deep currently expects zero
2398 collisions, so your algorithm has to be I<perfect>, so to speak.
2399 Collision detection may be introduced in a later version.
2403 You can specify a custom digest algorithm by calling the static C<set_digest()>
2404 function, passing a reference to a subroutine, and the length of the algorithm's
2405 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2406 objects. Here is a working example that uses a 256-bit hash from the
2407 I<Digest::SHA256> module. Please see
2408 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2413 my $context = Digest::SHA256::new(256);
2415 DBM::Deep::set_digest( \&my_digest, 32 );
2417 my $db = DBM::Deep->new( "foo-sha.db" );
2419 $db->{key1} = "value1";
2420 $db->{key2} = "value2";
2421 print "key1: " . $db->{key1} . "\n";
2422 print "key2: " . $db->{key2} . "\n";
2428 return substr( $context->hash($_[0]), 0, 32 );
2431 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2432 of bytes you specify in the C<set_digest()> function (in this case 32).
2434 =head1 CIRCULAR REFERENCES
2436 DBM::Deep has B<experimental> support for circular references. Meaning you
2437 can have a nested hash key or array element that points to a parent object.
2438 This relationship is stored in the DB file, and is preserved between sessions.
2441 my $db = DBM::Deep->new( "foo.db" );
2444 $db->{circle} = $db; # ref to self
2446 print $db->{foo} . "\n"; # prints "foo"
2447 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2449 One catch is, passing the object to a function that recursively walks the
2450 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2451 C<export()> methods) will result in an infinite loop. The other catch is,
2452 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2453 or C<next_key()> methods), you will get the I<target object's key>, not the
2454 ref's key. This gets even more interesting with the above example, where
2455 the I<circle> key points to the base DB object, which technically doesn't
2456 have a key. So I made DBM::Deep return "[base]" as the key name in that
2459 =head1 CAVEATS / ISSUES / BUGS
2461 This section describes all the known issues with DBM::Deep. It you have found
2462 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2464 =head2 UNUSED SPACE RECOVERY
2466 One major caveat with DBM::Deep is that space occupied by existing keys and
2467 values is not recovered when they are deleted. Meaning if you keep deleting
2468 and adding new keys, your file will continuously grow. I am working on this,
2469 but in the meantime you can call the built-in C<optimize()> method from time to
2470 time (perhaps in a crontab or something) to recover all your unused space.
2472 $db->optimize(); # returns true on success
2474 This rebuilds the ENTIRE database into a new file, then moves it on top of
2475 the original. The new file will have no unused space, thus it will take up as
2476 little disk space as possible. Please note that this operation can take
2477 a long time for large files, and you need enough disk space to temporarily hold
2478 2 copies of your DB file. The temporary file is created in the same directory
2479 as the original, named with a ".tmp" extension, and is deleted when the
2480 operation completes. Oh, and if locking is enabled, the DB is automatically
2481 locked for the entire duration of the copy.
2483 B<WARNING:> Only call optimize() on the top-level node of the database, and
2484 make sure there are no child references lying around. DBM::Deep keeps a reference
2485 counter, and if it is greater than 1, optimize() will abort and return undef.
2487 =head2 FILE CORRUPTION
2489 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2490 for a 32-bit signature when opened, but other corruption in files can cause
2491 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2492 stuck in an infinite loop depending on the level of corruption. File write
2493 operations are not checked for failure (for speed), so if you happen to run
2494 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2495 be addressed in a later version of DBM::Deep.
2499 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2500 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2501 about setting up your NFS server with a locking daemon, then using lockf() to
2502 lock your files, but your mileage may vary there as well. From what I
2503 understand, there is no real way to do it. However, if you need access to the
2504 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
2505 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2507 =head2 COPYING OBJECTS
2509 Beware of copying tied objects in Perl. Very strange things can happen.
2510 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2511 returns a new, blessed, tied hash or array to the same level in the DB.
2513 my $copy = $db->clone();
2515 B<Note>: Since clone() here is cloning the object, not the database location, any
2516 modifications to either $db or $copy will be visible in both.
2520 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2521 These functions cause every element in the array to move, which can be murder
2522 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2523 a different location. This will be addressed in the forthcoming version 1.00.
2525 =head2 WRITEONLY FILES
2527 If you pass in a filehandle to new(), you may have opened it in either a readonly or
2528 writeonly mode. STORE will verify that the filehandle is writable. However, there
2529 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
2530 filehandle isn't readable, it's not clear what will happen. So, don't do that.
2534 This section discusses DBM::Deep's speed and memory usage.
2538 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2539 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2540 multi-level hash/array support, and cross-platform FTPable files. Even so,
2541 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2542 with huge databases. Here is some test data:
2544 Adding 1,000,000 keys to new DB file...
2546 At 100 keys, avg. speed is 2,703 keys/sec
2547 At 200 keys, avg. speed is 2,642 keys/sec
2548 At 300 keys, avg. speed is 2,598 keys/sec
2549 At 400 keys, avg. speed is 2,578 keys/sec
2550 At 500 keys, avg. speed is 2,722 keys/sec
2551 At 600 keys, avg. speed is 2,628 keys/sec
2552 At 700 keys, avg. speed is 2,700 keys/sec
2553 At 800 keys, avg. speed is 2,607 keys/sec
2554 At 900 keys, avg. speed is 2,190 keys/sec
2555 At 1,000 keys, avg. speed is 2,570 keys/sec
2556 At 2,000 keys, avg. speed is 2,417 keys/sec
2557 At 3,000 keys, avg. speed is 1,982 keys/sec
2558 At 4,000 keys, avg. speed is 1,568 keys/sec
2559 At 5,000 keys, avg. speed is 1,533 keys/sec
2560 At 6,000 keys, avg. speed is 1,787 keys/sec
2561 At 7,000 keys, avg. speed is 1,977 keys/sec
2562 At 8,000 keys, avg. speed is 2,028 keys/sec
2563 At 9,000 keys, avg. speed is 2,077 keys/sec
2564 At 10,000 keys, avg. speed is 2,031 keys/sec
2565 At 20,000 keys, avg. speed is 1,970 keys/sec
2566 At 30,000 keys, avg. speed is 2,050 keys/sec
2567 At 40,000 keys, avg. speed is 2,073 keys/sec
2568 At 50,000 keys, avg. speed is 1,973 keys/sec
2569 At 60,000 keys, avg. speed is 1,914 keys/sec
2570 At 70,000 keys, avg. speed is 2,091 keys/sec
2571 At 80,000 keys, avg. speed is 2,103 keys/sec
2572 At 90,000 keys, avg. speed is 1,886 keys/sec
2573 At 100,000 keys, avg. speed is 1,970 keys/sec
2574 At 200,000 keys, avg. speed is 2,053 keys/sec
2575 At 300,000 keys, avg. speed is 1,697 keys/sec
2576 At 400,000 keys, avg. speed is 1,838 keys/sec
2577 At 500,000 keys, avg. speed is 1,941 keys/sec
2578 At 600,000 keys, avg. speed is 1,930 keys/sec
2579 At 700,000 keys, avg. speed is 1,735 keys/sec
2580 At 800,000 keys, avg. speed is 1,795 keys/sec
2581 At 900,000 keys, avg. speed is 1,221 keys/sec
2582 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2584 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2585 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2586 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2587 Run time was 12 min 3 sec.
2591 One of the great things about DBM::Deep is that it uses very little memory.
2592 Even with huge databases (1,000,000+ keys) you will not see much increased
2593 memory on your process. DBM::Deep relies solely on the filesystem for storing
2594 and fetching data. Here is output from I</usr/bin/top> before even opening a
2597 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2598 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2600 Basically the process is taking 2,716K of memory. And here is the same
2601 process after storing and fetching 1,000,000 keys:
2603 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2604 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2606 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2607 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2609 =head1 DB FILE FORMAT
2611 In case you were interested in the underlying DB file format, it is documented
2612 here in this section. You don't need to know this to use the module, it's just
2613 included for reference.
2617 DBM::Deep files always start with a 32-bit signature to identify the file type.
2618 This is at offset 0. The signature is "DPDB" in network byte order. This is
2619 checked for when the file is opened and an error will be thrown if it's not found.
2623 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2624 has a standard header containing the type of data, the length of data, and then
2625 the data itself. The type is a single character (1 byte), the length is a
2626 32-bit unsigned long in network byte order, and the data is, well, the data.
2627 Here is how it unfolds:
2631 Immediately after the 32-bit file signature is the I<Master Index> record.
2632 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2633 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2634 depending on how the DBM::Deep object was constructed.
2636 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2637 number). The first 8-bit char of the MD5 signature is the offset into the
2638 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2639 index element is a file offset of the next tag for the key/element in question,
2640 which is usually a I<Bucket List> tag (see below).
2642 The next tag I<could> be another index, depending on how many keys/elements
2643 exist. See L<RE-INDEXING> below for details.
2647 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2648 file offsets to where the actual data is stored. It starts with a standard
2649 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2650 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2651 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2652 When the list fills up, a I<Re-Index> operation is performed (See
2653 L<RE-INDEXING> below).
2657 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2658 index/value pair (in array mode). It starts with a standard tag header with
2659 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2660 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2661 header. The size reported in the tag header is only for the value, but then,
2662 just after the value is another size (32-bit unsigned long) and then the plain
2663 key itself. Since the value is likely to be fetched more often than the plain
2664 key, I figured it would be I<slightly> faster to store the value first.
2666 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2667 record for the nested structure, where the process begins all over again.
2671 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2672 exhausted. Then, when another key/element comes in, the list is converted to a
2673 new index record. However, this index will look at the next char in the MD5
2674 hash, and arrange new Bucket List pointers accordingly. This process is called
2675 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2676 17 (16 + new one) keys/elements are removed from the old Bucket List and
2677 inserted into the new index. Several new Bucket Lists are created in the
2678 process, as a new MD5 char from the key is being examined (it is unlikely that
2679 the keys will all share the same next char of their MD5s).
2681 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2682 when the Bucket Lists will turn into indexes, but the first round tends to
2683 happen right around 4,000 keys. You will see a I<slight> decrease in
2684 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2685 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2686 right around 900,000 keys. This process can continue nearly indefinitely --
2687 right up until the point the I<MD5> signatures start colliding with each other,
2688 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2689 getting struck by lightning while you are walking to cash in your tickets.
2690 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2691 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2692 this is 340 unodecillion, but don't quote me).
2696 When a new key/element is stored, the key (or index number) is first run through
2697 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2698 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2699 for the first char of the signature (in this case I<b0>). If it does not exist,
2700 a new I<Bucket List> is created for our key (and the next 15 future keys that
2701 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2702 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2703 this point, unless we are replacing an existing I<Bucket>), where the actual
2704 data will be stored.
2708 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2709 (or index number), then walking along the indexes. If there are enough
2710 keys/elements in this DB level, there might be nested indexes, each linked to
2711 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2712 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2713 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2714 plain key are stored.
2716 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2717 methods. In this process the indexes are walked systematically, and each key
2718 fetched in increasing MD5 order (which is why it appears random). Once the
2719 I<Bucket> is found, the value is skipped and the plain key returned instead.
2720 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2721 alphabetically sorted. This only happens on an index-level -- as soon as the
2722 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2723 so it's pretty much undefined how the keys will come out -- just like Perl's
2726 =head1 CODE COVERAGE
2728 We use B<Devel::Cover> to test the code coverage of our tests, below is the
2729 B<Devel::Cover> report on this module's test suite.
2731 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2732 File stmt bran cond sub pod time total
2733 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2734 blib/lib/DBM/Deep.pm 95.4 84.6 69.1 98.2 100.0 60.3 91.0
2735 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 n/a 26.4 98.0
2736 blib/lib/DBM/Deep/Hash.pm 95.3 80.0 100.0 100.0 n/a 13.3 92.4
2737 Total 96.4 85.4 73.1 98.8 100.0 100.0 92.4
2738 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2740 =head1 MORE INFORMATION
2742 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
2743 or send email to L<DBM-Deep@googlegroups.com>.
2747 Joseph Huckaby, L<jhuckaby@cpan.org>
2749 Rob Kinyon, L<rkinyon@cpan.org>
2751 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2755 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2756 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2760 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
2761 This is free software, you may use it and distribute it under the
2762 same terms as Perl itself.