7 # Multi-level database module for storing hash trees, arrays and simple
8 # key/value pairs into FTP-able, cross-platform binary database files.
10 # Type `perldoc DBM::Deep` for complete documentation.
14 # tie %db, 'DBM::Deep', 'my_database.db'; # standard tie() method
16 # my $db = new DBM::Deep( 'my_database.db' ); # preferred OO method
18 # $db->{my_scalar} = 'hello world';
19 # $db->{my_hash} = { larry => 'genius', hashes => 'fast' };
20 # $db->{my_array} = [ 1, 2, 3, time() ];
21 # $db->{my_complex} = [ 'hello', { perl => 'rules' }, 42, 99 ];
22 # push @{$db->{my_array}}, 'another value';
23 # my @key_list = keys %{$db->{my_hash}};
24 # print "This module " . $db->{my_complex}->[1]->{perl} . "!\n";
27 # (c) 2002-2005 Joseph Huckaby. All Rights Reserved.
28 # This program is free software; you can redistribute it and/or
29 # modify it under the same terms as Perl itself.
34 use Fcntl qw( :DEFAULT :flock :seek );
38 use vars qw( $VERSION );
42 # Set to 4 and 'N' for 32-bit offset tags (default). Theoretical limit of 4 GB per file.
43 # (Perl must be compiled with largefile support for files > 2 GB)
45 # Set to 8 and 'Q' for 64-bit offsets. Theoretical limit of 16 XB per file.
46 # (Perl must be compiled with largefile and 64-bit long support)
52 # Set to 4 and 'N' for 32-bit data length prefixes. Limit of 4 GB for each key/value.
53 # Upgrading this is possible (see above) but probably not necessary. If you need
54 # more than 4 GB for a single key or value, this module is really not for you :-)
56 #my $DATA_LENGTH_SIZE = 4;
57 #my $DATA_LENGTH_PACK = 'N';
58 our ($LONG_SIZE, $LONG_PACK, $DATA_LENGTH_SIZE, $DATA_LENGTH_PACK);
61 # Maximum number of buckets per list before another level of indexing is done.
62 # Increase this value for slightly greater speed, but larger database files.
63 # DO NOT decrease this value below 16, due to risk of recursive reindex overrun.
68 # Better not adjust anything below here, unless you're me :-)
72 # Setup digest function for keys
74 our ($DIGEST_FUNC, $HASH_SIZE);
75 #my $DIGEST_FUNC = \&Digest::MD5::md5;
78 # Precalculate index and bucket sizes based on values above.
81 my ($INDEX_SIZE, $BUCKET_SIZE, $BUCKET_LIST_SIZE);
88 # Setup file and tag signatures. These should never change.
90 sub SIG_FILE () { 'DPDB' }
91 sub SIG_HASH () { 'H' }
92 sub SIG_ARRAY () { 'A' }
93 sub SIG_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 () { return SIG_HASH; }
103 sub TYPE_ARRAY () { return SIG_ARRAY; }
105 sub _get_self { $_[0] }
108 # Class constructor method for Perl OO interface.
109 # Calls tie() and returns blessed reference to tied hash or array,
110 # providing a hybrid OO/tie interface.
114 if (scalar(@_) > 1) { $args = {@_}; }
115 else { $args = { file => shift }; }
118 # Check if we want a tied hash or array.
121 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
122 $class = 'DBM::Deep::Array';
123 require DBM::Deep::Array;
124 tie @$self, $class, %$args;
127 $class = 'DBM::Deep::Hash';
128 require DBM::Deep::Hash;
129 tie %$self, $class, %$args;
132 return bless $self, $class;
136 my @outer_params = qw( type base_offset );
139 # Setup $self and bless into this class.
146 base_offset => length(SIG_FILE),
151 foreach my $outer_parm ( @outer_params ) {
152 next unless exists $args->{$outer_parm};
153 $self->{$outer_parm} = delete $args->{$outer_parm}
156 $self->{root} = exists $args->{root}
158 : DBM::Deep::_::Root->new( $args );
160 if (!defined($self->fh)) { $self->_open(); }
168 require DBM::Deep::Hash;
169 return DBM::Deep::Hash->TIEHASH( @_ );
174 require DBM::Deep::Array;
175 return DBM::Deep::Array->TIEARRAY( @_ );
178 #XXX Unneeded now ...
184 # Open a FileHandle to the database, create if nonexistent.
185 # Make sure file signature matches DeepDB spec.
187 my $self = $_[0]->_get_self;#_get_self($_[0]);
189 if (defined($self->fh)) { $self->_close(); }
192 # Theoretically, adding O_BINARY should remove the need for the binmode
193 # Of course, testing it is going to be ... interesting.
194 my $flags = O_RDWR | O_CREAT | O_BINARY;
196 #XXX Can the mode be anything but r+, w+, or a+??
197 #XXX ie, it has to be in read-write mode
198 #XXX So, should we verify that the mode is legitimate?
200 #XXX Maybe the mode thingy should just go away. There's no good
201 #XXX reason for it ...
202 if ( $self->root->{mode} eq 'w+' ) {
207 sysopen( $fh, $self->root->{file}, $flags )
209 $self->root->{fh} = $fh;
210 }; if ($@ ) { $self->_throw_error( "Received error: $@\n" ); }
211 if (! defined($self->fh)) {
212 return $self->_throw_error("Cannot sysopen file: " . $self->root->{file} . ": $!");
217 #XXX Can we remove this by using the right sysopen() flags?
218 # Maybe ... q.v. above
219 binmode $fh; # for win32
221 if ($self->root->{autoflush}) {
222 my $old = select $fh;
228 seek($fh, 0, SEEK_SET);
231 my $bytes_read = read( $fh, $signature, length(SIG_FILE));
234 # File is empty -- write signature and master index
237 seek($fh, 0, SEEK_SET);
239 $self->root->{end} = length(SIG_FILE);
240 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
242 my $plain_key = "[base]";
243 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
244 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
246 # Flush the filehandle
247 my $old_fh = select $fh;
257 # Check signature was valid
259 unless ($signature eq SIG_FILE) {
261 return $self->_throw_error("Signature not found -- file is not a Deep DB");
264 $self->root->{end} = (stat($fh))[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
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");
285 # Close database FileHandle
287 my $self = $_[0]->_get_self;#_get_self($_[0]);
288 close $self->root->{fh};
293 # Given offset, signature and content, create tag and write to disk
295 my ($self, $offset, $sig, $content) = @_;
296 my $size = length($content);
300 seek($fh, $offset, SEEK_SET);
301 print($fh $sig . pack($DATA_LENGTH_PACK, $size) . $content );
303 if ($offset == $self->root->{end}) {
304 $self->root->{end} += SIG_SIZE + $DATA_LENGTH_SIZE + $size;
310 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
317 # Given offset, load single tag and return signature, size and data
324 seek($fh, $offset, SEEK_SET);
325 if (eof $fh) { return undef; }
328 read( $fh, $sig, SIG_SIZE);
331 read( $fh, $size, $DATA_LENGTH_SIZE);
332 $size = unpack($DATA_LENGTH_PACK, $size);
335 read( $fh, $buffer, $size);
340 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
347 # Given index tag, lookup single entry in index and return .
350 my ($tag, $index) = @_;
352 my $location = unpack($LONG_PACK, substr($tag->{content}, $index * $LONG_SIZE, $LONG_SIZE) );
353 if (!$location) { return; }
355 return $self->_load_tag( $location );
360 # Adds one key/value pair to bucket list, given offset, MD5 digest of key,
361 # plain (undigested) key and value.
364 my ($tag, $md5, $plain_key, $value) = @_;
365 my $keys = $tag->{content};
369 my $is_dbm_deep = eval { $value->isa( 'DBM::Deep' ) };
370 my $internal_ref = $is_dbm_deep && ($value->root eq $self->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 $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
379 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
382 # Found empty bucket (end of list). Populate and exit loop.
386 $location = $internal_ref
387 ? $value->base_offset
388 : $self->root->{end};
390 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
391 print($fh $md5 . pack($LONG_PACK, $location) );
394 elsif ($md5 eq $key) {
396 # Found existing bucket with same key. Replace with new value.
401 $location = $value->base_offset;
402 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
403 print($fh $md5 . pack($LONG_PACK, $location) );
406 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
408 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
411 # If value is a hash, array, or raw value with equal or less size, we can
412 # reuse the same content area of the database. Otherwise, we have to create
413 # a new content area at the EOF.
416 my $r = Scalar::Util::reftype( $value ) || '';
417 if ( $r eq 'HASH' || $r eq 'ARRAY' ) { $actual_length = $INDEX_SIZE; }
418 else { $actual_length = length($value); }
420 if ($actual_length <= $size) {
424 $location = $self->root->{end};
425 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE, SEEK_SET);
426 print($fh pack($LONG_PACK, $location) );
434 # If this is an internal reference, return now.
435 # No need to write value or plain key
442 # If bucket didn't fit into list, split into a new index level
445 seek($fh, $tag->{ref_loc}, SEEK_SET);
446 print($fh pack($LONG_PACK, $self->root->{end}) );
448 my $index_tag = $self->_create_tag($self->root->{end}, SIG_INDEX, chr(0) x $INDEX_SIZE);
451 $keys .= $md5 . pack($LONG_PACK, 0);
453 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
454 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
456 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
457 my $num = ord(substr($key, $tag->{ch} + 1, 1));
459 if ($offsets[$num]) {
460 my $offset = $offsets[$num] + SIG_SIZE + $DATA_LENGTH_SIZE;
461 seek($fh, $offset, SEEK_SET);
463 read( $fh, $subkeys, $BUCKET_LIST_SIZE);
465 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
466 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
468 seek($fh, $offset + ($k * $BUCKET_SIZE), SEEK_SET);
469 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
475 $offsets[$num] = $self->root->{end};
476 seek($fh, $index_tag->{offset} + ($num * $LONG_SIZE), SEEK_SET);
477 print($fh pack($LONG_PACK, $self->root->{end}) );
479 my $blist_tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
481 seek($fh, $blist_tag->{offset}, SEEK_SET);
482 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
487 $location ||= $self->root->{end};
488 } # re-index bucket list
491 # Seek to content area and store signature, value and plaintext key
495 seek($fh, $location, SEEK_SET);
498 # Write signature based on content type, set content length and write actual value.
500 my $r = Scalar::Util::reftype($value) || '';
502 print($fh TYPE_HASH );
503 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
504 $content_length = $INDEX_SIZE;
506 elsif ($r eq 'ARRAY') {
507 print($fh TYPE_ARRAY );
508 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
509 $content_length = $INDEX_SIZE;
511 elsif (!defined($value)) {
512 print($fh SIG_NULL );
513 print($fh pack($DATA_LENGTH_PACK, 0) );
517 print($fh SIG_DATA );
518 print($fh pack($DATA_LENGTH_PACK, length($value)) . $value );
519 $content_length = length($value);
523 # Plain key is stored AFTER value, as keys are typically fetched less often.
525 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
528 # If value is blessed, preserve class name
530 if ( $self->root->{autobless} ) {
531 my $value_class = Scalar::Util::blessed($value);
532 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
534 # Blessed ref -- will restore later
537 print($fh pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
538 $content_length += 1;
539 $content_length += $DATA_LENGTH_SIZE + length($value_class);
543 $content_length += 1;
548 # If this is a new content area, advance EOF counter
550 if ($location == $self->root->{end}) {
551 $self->root->{end} += SIG_SIZE;
552 $self->root->{end} += $DATA_LENGTH_SIZE + $content_length;
553 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
557 # If content is a hash or array, create new child DeepDB object and
558 # pass each key or element to it.
561 my $branch = DBM::Deep->new(
563 base_offset => $location,
566 foreach my $key (keys %{$value}) {
567 #$branch->{$key} = $value->{$key};
568 $branch->STORE( $key, $value->{$key} );
571 elsif ($r eq 'ARRAY') {
572 my $branch = DBM::Deep->new(
574 base_offset => $location,
578 foreach my $element (@{$value}) {
579 #$branch->[$index] = $element;
580 $branch->STORE( $index, $element );
588 return $self->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
591 sub _get_bucket_value {
593 # Fetch single value given tag and MD5 digested key.
596 my ($tag, $md5) = @_;
597 my $keys = $tag->{content};
602 # Iterate through buckets, looking for a key match
605 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
606 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
607 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
611 # Hit end of list, no match
616 if ( $md5 ne $key ) {
621 # Found match -- seek to offset and read signature
624 seek($fh, $subloc, SEEK_SET);
625 read( $fh, $signature, SIG_SIZE);
628 # If value is a hash or array, return new DeepDB object with correct offset
630 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
631 my $obj = DBM::Deep->new(
633 base_offset => $subloc,
637 if ($self->root->{autobless}) {
639 # Skip over value and plain key to see if object needs
642 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
645 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
646 if ($size) { seek($fh, $size, SEEK_CUR); }
649 read( $fh, $bless_bit, 1);
650 if (ord($bless_bit)) {
652 # Yes, object needs to be re-blessed
655 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
656 if ($size) { read( $fh, $class_name, $size); }
657 if ($class_name) { $obj = bless( $obj, $class_name ); }
665 # Otherwise return actual value
667 elsif ($signature eq SIG_DATA) {
670 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
671 if ($size) { read( $fh, $value, $size); }
676 # Key exists, but content is null
686 # Delete single key/value pair given tag and MD5 digested key.
689 my ($tag, $md5) = @_;
690 my $keys = $tag->{content};
695 # Iterate through buckets, looking for a key match
698 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
699 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
700 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
704 # Hit end of list, no match
709 if ( $md5 ne $key ) {
714 # Matched key -- delete bucket and return
716 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
717 print($fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
718 print($fh chr(0) x $BUCKET_SIZE );
728 # Check existence of single key given tag and MD5 digested key.
731 my ($tag, $md5) = @_;
732 my $keys = $tag->{content};
735 # Iterate through buckets, looking for a key match
738 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
739 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
740 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
744 # Hit end of list, no match
749 if ( $md5 ne $key ) {
754 # Matched key -- return true
762 sub _find_bucket_list {
764 # Locate offset for bucket list, given digested key
770 # Locate offset for bucket list using digest index system
773 my $tag = $self->_load_tag($self->base_offset);
774 if (!$tag) { return; }
776 while ($tag->{signature} ne SIG_BLIST) {
777 $tag = $self->_index_lookup($tag, ord(substr($md5, $ch, 1)));
778 if (!$tag) { return; }
785 sub _traverse_index {
787 # Scan index and recursively step into deeper levels, looking for next key.
789 my ($self, $offset, $ch, $force_return_next) = @_;
790 $force_return_next = undef unless $force_return_next;
792 my $tag = $self->_load_tag( $offset );
796 if ($tag->{signature} ne SIG_BLIST) {
797 my $content = $tag->{content};
799 if ($self->{return_next}) { $start = 0; }
800 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
802 for (my $index = $start; $index < 256; $index++) {
803 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
805 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
806 if (defined($result)) { return $result; }
810 $self->{return_next} = 1;
813 elsif ($tag->{signature} eq SIG_BLIST) {
814 my $keys = $tag->{content};
815 if ($force_return_next) { $self->{return_next} = 1; }
818 # Iterate through buckets, looking for a key match
820 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
821 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
822 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
826 # End of bucket list -- return to outer loop
828 $self->{return_next} = 1;
831 elsif ($key eq $self->{prev_md5}) {
833 # Located previous key -- return next one found
835 $self->{return_next} = 1;
838 elsif ($self->{return_next}) {
840 # Seek to bucket location and skip over signature
842 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
845 # Skip over value to get to plain key
848 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
849 if ($size) { seek($fh, $size, SEEK_CUR); }
852 # Read in plain key and return as scalar
855 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
856 if ($size) { read( $fh, $plain_key, $size); }
862 $self->{return_next} = 1;
863 } # tag is a bucket list
870 # Locate next key, given digested previous one
872 my $self = $_[0]->_get_self;#_get_self($_[0]);
874 $self->{prev_md5} = $_[1] ? $_[1] : undef;
875 $self->{return_next} = 0;
878 # If the previous key was not specifed, start at the top and
879 # return the first one found.
881 if (!$self->{prev_md5}) {
882 $self->{prev_md5} = chr(0) x $HASH_SIZE;
883 $self->{return_next} = 1;
886 return $self->_traverse_index( $self->base_offset, 0 );
891 # If db locking is set, flock() the db file. If called multiple
892 # times before unlock(), then the same number of unlocks() must
893 # be called before the lock is released.
895 my $self = $_[0]->_get_self;#_get_self($_[0]);
897 $type = LOCK_EX unless defined $type;
899 if ($self->root->{locking}) {
900 if (!$self->root->{locked}) { flock($self->fh, $type); }
901 $self->root->{locked}++;
911 # If db locking is set, unlock the db file. See note in lock()
912 # regarding calling lock() multiple times.
914 my $self = $_[0]->_get_self;#_get_self($_[0]);
916 if ($self->root->{locking} && $self->root->{locked} > 0) {
917 $self->root->{locked}--;
918 if (!$self->root->{locked}) { flock($self->fh, LOCK_UN); }
926 #XXX These uses of ref() need verified
929 # Copy single level of keys or elements to new DB handle.
930 # Recurse for nested structures
932 my $self = $_[0]->_get_self;#_get_self($_[0]);
935 if ($self->type eq TYPE_HASH) {
936 my $key = $self->first_key();
938 my $value = $self->get($key);
939 #XXX This doesn't work with autobless
940 if (!ref($value)) { $db_temp->{$key} = $value; }
942 my $type = $value->type;
943 if ($type eq TYPE_HASH) { $db_temp->{$key} = {}; }
944 else { $db_temp->{$key} = []; }
945 $value->_copy_node( $db_temp->{$key} );
947 $key = $self->next_key($key);
951 my $length = $self->length();
952 for (my $index = 0; $index < $length; $index++) {
953 my $value = $self->get($index);
954 if (!ref($value)) { $db_temp->[$index] = $value; }
955 #XXX NO tests for this code
957 my $type = $value->type;
958 if ($type eq TYPE_HASH) { $db_temp->[$index] = {}; }
959 else { $db_temp->[$index] = []; }
960 $value->_copy_node( $db_temp->[$index] );
968 # Recursively export into standard Perl hashes and arrays.
970 my $self = $_[0]->_get_self;#_get_self($_[0]);
973 if ($self->type eq TYPE_HASH) { $temp = {}; }
974 elsif ($self->type eq TYPE_ARRAY) { $temp = []; }
977 $self->_copy_node( $temp );
985 # Recursively import Perl hash/array structure
987 #XXX This use of ref() seems to be ok
988 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
990 my $self = $_[0]->_get_self;#_get_self($_[0]);
993 #XXX This use of ref() seems to be ok
996 # struct is not a reference, so just import based on our type
1000 if ($self->type eq TYPE_HASH) { $struct = {@_}; }
1001 elsif ($self->type eq TYPE_ARRAY) { $struct = [@_]; }
1004 my $r = Scalar::Util::reftype($struct) || '';
1005 if ($r eq "HASH" && $self->type eq TYPE_HASH) {
1006 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
1008 elsif ($r eq "ARRAY" && $self->type eq TYPE_ARRAY) {
1009 $self->push( @$struct );
1012 return $self->_throw_error("Cannot import: type mismatch");
1020 # Rebuild entire database into new file, then move
1021 # it back on top of original.
1023 my $self = $_[0]->_get_self;#_get_self($_[0]);
1025 #XXX Need to create a new test for this
1026 # if ($self->root->{links} > 1) {
1027 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
1030 my $db_temp = DBM::Deep->new(
1031 file => $self->root->{file} . '.tmp',
1035 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
1039 $self->_copy_node( $db_temp );
1043 # Attempt to copy user, group and permissions over to new file
1045 my @stats = stat($self->fh);
1046 my $perms = $stats[2] & 07777;
1047 my $uid = $stats[4];
1048 my $gid = $stats[5];
1049 chown( $uid, $gid, $self->root->{file} . '.tmp' );
1050 chmod( $perms, $self->root->{file} . '.tmp' );
1052 # q.v. perlport for more information on this variable
1053 if ( $^O eq 'MSWin32' ) {
1055 # Potential race condition when optmizing on Win32 with locking.
1056 # The Windows filesystem requires that the filehandle be closed
1057 # before it is overwritten with rename(). This could be redone
1064 if (!rename $self->root->{file} . '.tmp', $self->root->{file}) {
1065 unlink $self->root->{file} . '.tmp';
1067 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
1079 # Make copy of object and return
1081 my $self = $_[0]->_get_self;#_get_self($_[0]);
1083 return DBM::Deep->new(
1084 type => $self->type,
1085 base_offset => $self->base_offset,
1091 my %is_legal_filter = map {
1094 store_key store_value
1095 fetch_key fetch_value
1100 # Setup filter function for storing or fetching the key or value
1102 my $self = $_[0]->_get_self;#_get_self($_[0]);
1103 my $type = lc $_[1];
1104 my $func = $_[2] ? $_[2] : undef;
1106 if ( $is_legal_filter{$type} ) {
1107 $self->root->{"filter_$type"} = $func;
1121 # Get access to the root structure
1123 my $self = $_[0]->_get_self;#_get_self($_[0]);
1124 return $self->{root};
1129 # Get access to the raw FileHandle
1131 #XXX It will be useful, though, when we split out HASH and ARRAY
1132 my $self = $_[0]->_get_self;#_get_self($_[0]);
1133 return $self->root->{fh};
1138 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
1140 my $self = $_[0]->_get_self;#_get_self($_[0]);
1141 return $self->{type};
1146 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
1148 my $self = $_[0]->_get_self;#_get_self($_[0]);
1149 return $self->{base_offset};
1154 # Get last error string, or undef if no error
1157 #? ( _get_self($_[0])->{root}->{error} or undef )
1158 ? ( $_[0]->_get_self->{root}->{error} or undef )
1168 # Store error string in self
1170 my $self = $_[0]->_get_self;#_get_self($_[0]);
1171 my $error_text = $_[1];
1173 $self->root->{error} = $error_text;
1175 unless ($self->root->{debug}) {
1176 die "DBM::Deep: $error_text\n";
1179 warn "DBM::Deep: $error_text\n";
1187 my $self = $_[0]->_get_self;#_get_self($_[0]);
1189 undef $self->root->{error};
1194 # Precalculate index, bucket and bucket list sizes
1197 #XXX I don't like this ...
1198 set_pack() unless defined $LONG_SIZE;
1200 $INDEX_SIZE = 256 * $LONG_SIZE;
1201 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
1202 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
1207 # Set pack/unpack modes (see file header for more)
1209 my ($long_s, $long_p, $data_s, $data_p) = @_;
1211 $LONG_SIZE = $long_s ? $long_s : 4;
1212 $LONG_PACK = $long_p ? $long_p : 'N';
1214 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
1215 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
1222 # Set key digest function (default is MD5)
1224 my ($digest_func, $hash_size) = @_;
1226 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
1227 $HASH_SIZE = $hash_size ? $hash_size : 16;
1233 # tie() methods (hashes and arrays)
1238 # Store single hash key/value or array element in database.
1240 my $self = $_[0]->_get_self;#_get_self($_[0]);
1241 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1242 #XXX What is ref() checking here?
1243 #YYY User may be storing a hash, in which case we do not want it run
1244 #YYY through the filtering system
1245 my $value = ($self->root->{filter_store_value} && !ref($_[2])) ? $self->root->{filter_store_value}->($_[2]) : $_[2];
1247 my $unpacked_key = $key;
1248 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1249 my $md5 = $DIGEST_FUNC->($key);
1252 # Make sure file is open
1254 if (!defined($self->fh) && !$self->_open()) {
1262 # Request exclusive lock for writing
1264 $self->lock( LOCK_EX );
1267 # If locking is enabled, set 'end' parameter again, in case another
1268 # DB instance appended to our file while we were unlocked.
1270 if ($self->root->{locking} || $self->root->{volatile}) {
1271 $self->root->{end} = (stat($fh))[7];
1275 # Locate offset for bucket list using digest index system
1277 my $tag = $self->_load_tag($self->base_offset);
1279 $tag = $self->_create_tag($self->base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1283 while ($tag->{signature} ne SIG_BLIST) {
1284 my $num = ord(substr($md5, $ch, 1));
1285 my $new_tag = $self->_index_lookup($tag, $num);
1287 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1288 seek($fh, $ref_loc, SEEK_SET);
1289 print($fh pack($LONG_PACK, $self->root->{end}) );
1291 $tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1292 $tag->{ref_loc} = $ref_loc;
1297 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1299 $tag->{ref_loc} = $ref_loc;
1306 # Add key/value to bucket list
1308 my $result = $self->_add_bucket( $tag, $md5, $key, $value );
1311 # If this object is an array, and bucket was not a replace, and key is numerical,
1312 # and index is equal or greater than current length, advance length variable.
1314 if (($result == 2) && ($self->type eq TYPE_ARRAY) && ($unpacked_key =~ /^\d+$/) && ($unpacked_key >= $self->FETCHSIZE())) {
1315 $self->STORESIZE( $unpacked_key + 1 );
1325 # Fetch single value or element given plain key or array index
1327 my $self = $_[0]->_get_self;#_get_self($_[0]);
1330 if ( $self->type eq TYPE_HASH ) {
1331 if ( my $filter = $self->root->{filter_store_key} ) {
1332 $key = $filter->( $key );
1335 elsif ( $self->type eq TYPE_ARRAY ) {
1336 if ( $key =~ /^\d+$/ ) {
1337 $key = pack($LONG_PACK, $key);
1341 my $md5 = $DIGEST_FUNC->($key);
1344 # Make sure file is open
1346 if (!defined($self->fh)) { $self->_open(); }
1349 # Request shared lock for reading
1351 $self->lock( LOCK_SH );
1353 my $tag = $self->_find_bucket_list( $md5 );
1360 # Get value from bucket list
1362 my $result = $self->_get_bucket_value( $tag, $md5 );
1366 #XXX What is ref() checking here?
1367 return ($result && !ref($result) && $self->root->{filter_fetch_value}) ? $self->root->{filter_fetch_value}->($result) : $result;
1372 # Delete single key/value pair or element given plain key or array index
1374 my $self = $_[0]->_get_self;#_get_self($_[0]);
1375 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1377 my $unpacked_key = $key;
1378 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1379 my $md5 = $DIGEST_FUNC->($key);
1382 # Make sure file is open
1384 if (!defined($self->fh)) { $self->_open(); }
1387 # Request exclusive lock for writing
1389 $self->lock( LOCK_EX );
1391 my $tag = $self->_find_bucket_list( $md5 );
1400 my $result = $self->_delete_bucket( $tag, $md5 );
1403 # If this object is an array and the key deleted was on the end of the stack,
1404 # decrement the length variable.
1406 if ($result && ($self->type eq TYPE_ARRAY) && ($unpacked_key == $self->FETCHSIZE() - 1)) {
1407 $self->STORESIZE( $unpacked_key );
1417 # Check if a single key or element exists given plain key or array index
1419 my $self = $_[0]->_get_self;#_get_self($_[0]);
1420 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1422 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1423 my $md5 = $DIGEST_FUNC->($key);
1426 # Make sure file is open
1428 if (!defined($self->fh)) { $self->_open(); }
1431 # Request shared lock for reading
1433 $self->lock( LOCK_SH );
1435 my $tag = $self->_find_bucket_list( $md5 );
1438 # For some reason, the built-in exists() function returns '' for false
1446 # Check if bucket exists and return 1 or ''
1448 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1457 # Clear all keys from hash, or all elements from array.
1459 my $self = $_[0]->_get_self;#_get_self($_[0]);
1462 # Make sure file is open
1464 if (!defined($self->fh)) { $self->_open(); }
1467 # Request exclusive lock for writing
1469 $self->lock( LOCK_EX );
1473 seek($fh, $self->base_offset, SEEK_SET);
1479 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
1487 # Public method aliases
1489 *put = *store = *STORE;
1490 *get = *fetch = *FETCH;
1495 package DBM::Deep::_::Root;
1510 filter_store_key => undef,
1511 filter_store_value => undef,
1512 filter_fetch_key => undef,
1513 filter_fetch_value => undef,
1524 return unless $self;
1526 close $self->{fh} if $self->{fh};
1537 DBM::Deep - A pure perl multi-level hash/array DBM
1542 my $db = DBM::Deep->new( "foo.db" );
1544 $db->{key} = 'value'; # tie() style
1547 $db->put('key', 'value'); # OO style
1548 print $db->get('key');
1550 # true multi-level support
1551 $db->{my_complex} = [
1552 'hello', { perl => 'rules' },
1557 A unique flat-file database module, written in pure perl. True
1558 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1559 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1560 handle millions of keys and unlimited hash levels without significant
1561 slow-down. Written from the ground-up in pure perl -- this is NOT a
1562 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1563 Mac OS X and Windows.
1567 Hopefully you are using CPAN's excellent Perl module, which will download
1568 and install the module for you. If not, get the tarball, and run these
1580 Construction can be done OO-style (which is the recommended way), or using
1581 Perl's tie() function. Both are examined here.
1583 =head2 OO CONSTRUCTION
1585 The recommended way to construct a DBM::Deep object is to use the new()
1586 method, which gets you a blessed, tied hash or array reference.
1588 my $db = DBM::Deep->new( "foo.db" );
1590 This opens a new database handle, mapped to the file "foo.db". If this
1591 file does not exist, it will automatically be created. DB files are
1592 opened in "r+" (read/write) mode, and the type of object returned is a
1593 hash, unless otherwise specified (see L<OPTIONS> below).
1597 You can pass a number of options to the constructor to specify things like
1598 locking, autoflush, etc. This is done by passing an inline hash:
1600 my $db = DBM::Deep->new(
1606 Notice that the filename is now specified I<inside> the hash with
1607 the "file" parameter, as opposed to being the sole argument to the
1608 constructor. This is required if any options are specified.
1609 See L<OPTIONS> below for the complete list.
1613 You can also start with an array instead of a hash. For this, you must
1614 specify the C<type> parameter:
1616 my $db = DBM::Deep->new(
1618 type => DBM::Deep->TYPE_ARRAY
1621 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1622 a new DB file. If you create a DBM::Deep object with an existing file, the
1623 C<type> will be loaded from the file header, and ignored if it is passed
1626 =head2 TIE CONSTRUCTION
1628 Alternatively, you can create a DBM::Deep handle by using Perl's built-in
1629 tie() function. This is not ideal, because you get only a basic, tied hash
1630 (or array) which is not blessed, so you can't call any functions on it.
1633 tie %hash, "DBM::Deep", "foo.db";
1636 tie @array, "DBM::Deep", "bar.db";
1638 As with the OO constructor, you can replace the DB filename parameter with
1639 a hash containing one or more options (see L<OPTIONS> just below for the
1642 tie %hash, "DBM::Deep", {
1650 There are a number of options that can be passed in when constructing your
1651 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1657 Filename of the DB file to link the handle to. You can pass a full absolute
1658 filesystem path, partial path, or a plain filename if the file is in the
1659 current working directory. This is a required parameter.
1663 File open mode (read-only, read-write, etc.) string passed to Perl's FileHandle
1664 module. This is an optional parameter, and defaults to "r+" (read/write).
1665 B<Note:> If the default (r+) mode is selected, the file will also be auto-
1666 created if it doesn't exist.
1670 This parameter specifies what type of object to create, a hash or array. Use
1671 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1672 This only takes effect when beginning a new file. This is an optional
1673 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1677 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1678 function to lock the database in exclusive mode for writes, and shared mode for
1679 reads. Pass any true value to enable. This affects the base DB handle I<and
1680 any child hashes or arrays> that use the same DB file. This is an optional
1681 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1685 Specifies whether autoflush is to be enabled on the underlying FileHandle.
1686 This obviously slows down write operations, but is required if you may have
1687 multiple processes accessing the same DB file (also consider enable I<locking>
1688 or at least I<volatile>). Pass any true value to enable. This is an optional
1689 parameter, and defaults to 0 (disabled).
1693 If I<volatile> mode is enabled, DBM::Deep will stat() the DB file before each
1694 STORE() operation. This is required if an outside force may change the size of
1695 the file between transactions. Locking also implicitly enables volatile. This
1696 is useful if you want to use a different locking system or write your own. Pass
1697 any true value to enable. This is an optional parameter, and defaults to 0
1702 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1703 restore them when fetched. This is an B<experimental> feature, and does have
1704 side-effects. Basically, when hashes are re-blessed into their original
1705 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1706 able to call any DBM::Deep methods on them. You have been warned.
1707 This is an optional parameter, and defaults to 0 (disabled).
1711 See L<FILTERS> below.
1715 Setting I<debug> mode will make all errors non-fatal, dump them out to
1716 STDERR, and continue on. This is for debugging purposes only, and probably
1717 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1721 Instead of passing a file path, you can instead pass a handle to an pre-opened
1722 filehandle. Note: Beware of using the magick *DATA handle, as this actually
1723 contains your entire Perl script, as well as the data following the __DATA__
1724 marker. This will not work, because DBM::Deep uses absolute seek()s into the
1725 file. Instead, consider reading *DATA into an IO::Scalar handle, then passing
1730 =head1 TIE INTERFACE
1732 With DBM::Deep you can access your databases using Perl's standard hash/array
1733 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can treat
1734 them as such. DBM::Deep will intercept all reads/writes and direct them to the right
1735 place -- the DB file. This has nothing to do with the L<TIE CONSTRUCTION>
1736 section above. This simply tells you how to use DBM::Deep using regular hashes
1737 and arrays, rather than calling functions like C<get()> and C<put()> (although those
1738 work too). It is entirely up to you how to want to access your databases.
1742 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1743 or even nested hashes (or arrays) using standard Perl syntax:
1745 my $db = DBM::Deep->new( "foo.db" );
1747 $db->{mykey} = "myvalue";
1749 $db->{myhash}->{subkey} = "subvalue";
1751 print $db->{myhash}->{subkey} . "\n";
1753 You can even step through hash keys using the normal Perl C<keys()> function:
1755 foreach my $key (keys %$db) {
1756 print "$key: " . $db->{$key} . "\n";
1759 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1760 pushes them onto an array, all before the loop even begins. If you have an
1761 extra large hash, this may exhaust Perl's memory. Instead, consider using
1762 Perl's C<each()> function, which pulls keys/values one at a time, using very
1765 while (my ($key, $value) = each %$db) {
1766 print "$key: $value\n";
1769 Please note that when using C<each()>, you should always pass a direct
1770 hash reference, not a lookup. Meaning, you should B<never> do this:
1773 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1775 This causes an infinite loop, because for each iteration, Perl is calling
1776 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1777 it effectively keeps returning the first key over and over again. Instead,
1778 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1782 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1783 reference. This includes inserting, removing and manipulating elements,
1784 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1785 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1786 or simply be a nested array reference inside a hash. Example:
1788 my $db = DBM::Deep->new(
1789 file => "foo-array.db",
1790 type => DBM::Deep->TYPE_ARRAY
1794 push @$db, "bar", "baz";
1795 unshift @$db, "bah";
1797 my $last_elem = pop @$db; # baz
1798 my $first_elem = shift @$db; # bah
1799 my $second_elem = $db->[1]; # bar
1801 my $num_elements = scalar @$db;
1805 In addition to the I<tie()> interface, you can also use a standard OO interface
1806 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1807 array) has its own methods, but both types share the following common methods:
1808 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1814 Stores a new hash key/value pair, or sets an array element value. Takes two
1815 arguments, the hash key or array index, and the new value. The value can be
1816 a scalar, hash ref or array ref. Returns true on success, false on failure.
1818 $db->put("foo", "bar"); # for hashes
1819 $db->put(1, "bar"); # for arrays
1823 Fetches the value of a hash key or array element. Takes one argument: the hash
1824 key or array index. Returns a scalar, hash ref or array ref, depending on the
1827 my $value = $db->get("foo"); # for hashes
1828 my $value = $db->get(1); # for arrays
1832 Checks if a hash key or array index exists. Takes one argument: the hash key
1833 or array index. Returns true if it exists, false if not.
1835 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1836 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1840 Deletes one hash key/value pair or array element. Takes one argument: the hash
1841 key or array index. Returns true on success, false if not found. For arrays,
1842 the remaining elements located after the deleted element are NOT moved over.
1843 The deleted element is essentially just undefined, which is exactly how Perl's
1844 internal arrays work. Please note that the space occupied by the deleted
1845 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1846 below for details and workarounds.
1848 $db->delete("foo"); # for hashes
1849 $db->delete(1); # for arrays
1853 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1854 value. Please note that the space occupied by the deleted keys/values or
1855 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1856 details and workarounds.
1858 $db->clear(); # hashes or arrays
1864 For hashes, DBM::Deep supports all the common methods described above, and the
1865 following additional methods: C<first_key()> and C<next_key()>.
1871 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1872 fetched in an undefined order (which appears random). Takes no arguments,
1873 returns the key as a scalar value.
1875 my $key = $db->first_key();
1879 Returns the "next" key in the hash, given the previous one as the sole argument.
1880 Returns undef if there are no more keys to be fetched.
1882 $key = $db->next_key($key);
1886 Here are some examples of using hashes:
1888 my $db = DBM::Deep->new( "foo.db" );
1890 $db->put("foo", "bar");
1891 print "foo: " . $db->get("foo") . "\n";
1893 $db->put("baz", {}); # new child hash ref
1894 $db->get("baz")->put("buz", "biz");
1895 print "buz: " . $db->get("baz")->get("buz") . "\n";
1897 my $key = $db->first_key();
1899 print "$key: " . $db->get($key) . "\n";
1900 $key = $db->next_key($key);
1903 if ($db->exists("foo")) { $db->delete("foo"); }
1907 For arrays, DBM::Deep supports all the common methods described above, and the
1908 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1909 C<unshift()> and C<splice()>.
1915 Returns the number of elements in the array. Takes no arguments.
1917 my $len = $db->length();
1921 Adds one or more elements onto the end of the array. Accepts scalars, hash
1922 refs or array refs. No return value.
1924 $db->push("foo", "bar", {});
1928 Fetches the last element in the array, and deletes it. Takes no arguments.
1929 Returns undef if array is empty. Returns the element value.
1931 my $elem = $db->pop();
1935 Fetches the first element in the array, deletes it, then shifts all the
1936 remaining elements over to take up the space. Returns the element value. This
1937 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1940 my $elem = $db->shift();
1944 Inserts one or more elements onto the beginning of the array, shifting all
1945 existing elements over to make room. Accepts scalars, hash refs or array refs.
1946 No return value. This method is not recommended with large arrays -- see
1947 <LARGE ARRAYS> below for details.
1949 $db->unshift("foo", "bar", {});
1953 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1954 -f splice> for usage -- it is too complicated to document here. This method is
1955 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1959 Here are some examples of using arrays:
1961 my $db = DBM::Deep->new(
1963 type => DBM::Deep->TYPE_ARRAY
1966 $db->push("bar", "baz");
1967 $db->unshift("foo");
1970 my $len = $db->length();
1971 print "length: $len\n"; # 4
1973 for (my $k=0; $k<$len; $k++) {
1974 print "$k: " . $db->get($k) . "\n";
1977 $db->splice(1, 2, "biz", "baf");
1979 while (my $elem = shift @$db) {
1980 print "shifted: $elem\n";
1985 Enable automatic file locking by passing a true value to the C<locking>
1986 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1988 my $db = DBM::Deep->new(
1993 This causes DBM::Deep to C<flock()> the underlying FileHandle object with exclusive
1994 mode for writes, and shared mode for reads. This is required if you have
1995 multiple processes accessing the same database file, to avoid file corruption.
1996 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1997 NFS> below for more.
1999 =head2 EXPLICIT LOCKING
2001 You can explicitly lock a database, so it remains locked for multiple
2002 transactions. This is done by calling the C<lock()> method, and passing an
2003 optional lock mode argument (defaults to exclusive mode). This is particularly
2004 useful for things like counters, where the current value needs to be fetched,
2005 then incremented, then stored again.
2008 my $counter = $db->get("counter");
2010 $db->put("counter", $counter);
2019 You can pass C<lock()> an optional argument, which specifies which mode to use
2020 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2021 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2022 same as the constants defined in Perl's C<Fcntl> module.
2024 $db->lock( DBM::Deep->LOCK_SH );
2028 If you want to implement your own file locking scheme, be sure to create your
2029 DBM::Deep objects setting the C<volatile> option to true. This hints to DBM::Deep
2030 that the DB file may change between transactions. See L<LOW-LEVEL ACCESS>
2033 =head1 IMPORTING/EXPORTING
2035 You can import existing complex structures by calling the C<import()> method,
2036 and export an entire database into an in-memory structure using the C<export()>
2037 method. Both are examined here.
2041 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2042 walking the structure and adding keys/elements to the database as you go,
2043 simply pass a reference to the C<import()> method. This recursively adds
2044 everything to an existing DBM::Deep object for you. Here is an example:
2049 array1 => [ "elem0", "elem1", "elem2" ],
2051 subkey1 => "subvalue1",
2052 subkey2 => "subvalue2"
2056 my $db = DBM::Deep->new( "foo.db" );
2057 $db->import( $struct );
2059 print $db->{key1} . "\n"; # prints "value1"
2061 This recursively imports the entire C<$struct> object into C<$db>, including
2062 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2063 keys are merged with the existing ones, replacing if they already exist.
2064 The C<import()> method can be called on any database level (not just the base
2065 level), and works with both hash and array DB types.
2069 B<Note:> Make sure your existing structure has no circular references in it.
2070 These will cause an infinite loop when importing.
2074 Calling the C<export()> method on an existing DBM::Deep object will return
2075 a reference to a new in-memory copy of the database. The export is done
2076 recursively, so all nested hashes/arrays are all exported to standard Perl
2077 objects. Here is an example:
2079 my $db = DBM::Deep->new( "foo.db" );
2081 $db->{key1} = "value1";
2082 $db->{key2} = "value2";
2084 $db->{hash1}->{subkey1} = "subvalue1";
2085 $db->{hash1}->{subkey2} = "subvalue2";
2087 my $struct = $db->export();
2089 print $struct->{key1} . "\n"; # prints "value1"
2091 This makes a complete copy of the database in memory, and returns a reference
2092 to it. The C<export()> method can be called on any database level (not just
2093 the base level), and works with both hash and array DB types. Be careful of
2094 large databases -- you can store a lot more data in a DBM::Deep object than an
2095 in-memory Perl structure.
2099 B<Note:> Make sure your database has no circular references in it.
2100 These will cause an infinite loop when exporting.
2104 DBM::Deep has a number of hooks where you can specify your own Perl function
2105 to perform filtering on incoming or outgoing data. This is a perfect
2106 way to extend the engine, and implement things like real-time compression or
2107 encryption. Filtering applies to the base DB level, and all child hashes /
2108 arrays. Filter hooks can be specified when your DBM::Deep object is first
2109 constructed, or by calling the C<set_filter()> method at any time. There are
2110 four available filter hooks, described below:
2114 =item * filter_store_key
2116 This filter is called whenever a hash key is stored. It
2117 is passed the incoming key, and expected to return a transformed key.
2119 =item * filter_store_value
2121 This filter is called whenever a hash key or array element is stored. It
2122 is passed the incoming value, and expected to return a transformed value.
2124 =item * filter_fetch_key
2126 This filter is called whenever a hash key is fetched (i.e. via
2127 C<first_key()> or C<next_key()>). It is passed the transformed key,
2128 and expected to return the plain key.
2130 =item * filter_fetch_value
2132 This filter is called whenever a hash key or array element is fetched.
2133 It is passed the transformed value, and expected to return the plain value.
2137 Here are the two ways to setup a filter hook:
2139 my $db = DBM::Deep->new(
2141 filter_store_value => \&my_filter_store,
2142 filter_fetch_value => \&my_filter_fetch
2147 $db->set_filter( "filter_store_value", \&my_filter_store );
2148 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2150 Your filter function will be called only when dealing with SCALAR keys or
2151 values. When nested hashes and arrays are being stored/fetched, filtering
2152 is bypassed. Filters are called as static functions, passed a single SCALAR
2153 argument, and expected to return a single SCALAR value. If you want to
2154 remove a filter, set the function reference to C<undef>:
2156 $db->set_filter( "filter_store_value", undef );
2158 =head2 REAL-TIME ENCRYPTION EXAMPLE
2160 Here is a working example that uses the I<Crypt::Blowfish> module to
2161 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2162 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2163 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2166 use Crypt::Blowfish;
2169 my $cipher = Crypt::CBC->new({
2170 'key' => 'my secret key',
2171 'cipher' => 'Blowfish',
2173 'regenerate_key' => 0,
2174 'padding' => 'space',
2178 my $db = DBM::Deep->new(
2179 file => "foo-encrypt.db",
2180 filter_store_key => \&my_encrypt,
2181 filter_store_value => \&my_encrypt,
2182 filter_fetch_key => \&my_decrypt,
2183 filter_fetch_value => \&my_decrypt,
2186 $db->{key1} = "value1";
2187 $db->{key2} = "value2";
2188 print "key1: " . $db->{key1} . "\n";
2189 print "key2: " . $db->{key2} . "\n";
2195 return $cipher->encrypt( $_[0] );
2198 return $cipher->decrypt( $_[0] );
2201 =head2 REAL-TIME COMPRESSION EXAMPLE
2203 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2204 compression / decompression of keys & values with DBM::Deep Filters.
2205 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2206 more on I<Compress::Zlib>.
2211 my $db = DBM::Deep->new(
2212 file => "foo-compress.db",
2213 filter_store_key => \&my_compress,
2214 filter_store_value => \&my_compress,
2215 filter_fetch_key => \&my_decompress,
2216 filter_fetch_value => \&my_decompress,
2219 $db->{key1} = "value1";
2220 $db->{key2} = "value2";
2221 print "key1: " . $db->{key1} . "\n";
2222 print "key2: " . $db->{key2} . "\n";
2228 return Compress::Zlib::memGzip( $_[0] ) ;
2231 return Compress::Zlib::memGunzip( $_[0] ) ;
2234 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2235 actually numerical index numbers, and are not filtered.
2237 =head1 ERROR HANDLING
2239 Most DBM::Deep methods return a true value for success, and call die() on
2240 failure. You can wrap calls in an eval block to catch the die. Also, the
2241 actual error message is stored in an internal scalar, which can be fetched by
2242 calling the C<error()> method.
2244 my $db = DBM::Deep->new( "foo.db" ); # create hash
2245 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2247 print $db->error(); # prints error message
2249 You can then call C<clear_error()> to clear the current error state.
2253 If you set the C<debug> option to true when creating your DBM::Deep object,
2254 all errors are considered NON-FATAL, and dumped to STDERR. This is only
2255 for debugging purposes.
2257 =head1 LARGEFILE SUPPORT
2259 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2260 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2261 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2262 by calling the static C<set_pack()> method before you do anything else.
2264 DBM::Deep::set_pack(8, 'Q');
2266 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2267 instead of 32-bit longs. After setting these values your DB files have a
2268 theoretical maximum size of 16 XB (exabytes).
2272 B<Note:> Changing these values will B<NOT> work for existing database files.
2273 Only change this for new files, and make sure it stays set consistently
2274 throughout the file's life. If you do set these values, you can no longer
2275 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2276 back to 32-bit mode.
2280 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2281 only a 32-bit Perl. However, I have received user reports that this does
2284 =head1 LOW-LEVEL ACCESS
2286 If you require low-level access to the underlying FileHandle that DBM::Deep uses,
2287 you can call the C<fh()> method, which returns the handle:
2291 This method can be called on the root level of the datbase, or any child
2292 hashes or arrays. All levels share a I<root> structure, which contains things
2293 like the FileHandle, a reference counter, and all your options you specified
2294 when you created the object. You can get access to this root structure by
2295 calling the C<root()> method.
2297 my $root = $db->root();
2299 This is useful for changing options after the object has already been created,
2300 such as enabling/disabling locking, volatile or debug modes. You can also
2301 store your own temporary user data in this structure (be wary of name
2302 collision), which is then accessible from any child hash or array.
2304 =head1 CUSTOM DIGEST ALGORITHM
2306 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2307 keys. However you can override this, and use another algorithm (such as SHA-256)
2308 or even write your own. But please note that DBM::Deep currently expects zero
2309 collisions, so your algorithm has to be I<perfect>, so to speak.
2310 Collision detection may be introduced in a later version.
2314 You can specify a custom digest algorithm by calling the static C<set_digest()>
2315 function, passing a reference to a subroutine, and the length of the algorithm's
2316 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2317 objects. Here is a working example that uses a 256-bit hash from the
2318 I<Digest::SHA256> module. Please see
2319 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2324 my $context = Digest::SHA256::new(256);
2326 DBM::Deep::set_digest( \&my_digest, 32 );
2328 my $db = DBM::Deep->new( "foo-sha.db" );
2330 $db->{key1} = "value1";
2331 $db->{key2} = "value2";
2332 print "key1: " . $db->{key1} . "\n";
2333 print "key2: " . $db->{key2} . "\n";
2339 return substr( $context->hash($_[0]), 0, 32 );
2342 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2343 of bytes you specify in the C<set_digest()> function (in this case 32).
2345 =head1 CIRCULAR REFERENCES
2347 DBM::Deep has B<experimental> support for circular references. Meaning you
2348 can have a nested hash key or array element that points to a parent object.
2349 This relationship is stored in the DB file, and is preserved between sessions.
2352 my $db = DBM::Deep->new( "foo.db" );
2355 $db->{circle} = $db; # ref to self
2357 print $db->{foo} . "\n"; # prints "foo"
2358 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2360 One catch is, passing the object to a function that recursively walks the
2361 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2362 C<export()> methods) will result in an infinite loop. The other catch is,
2363 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2364 or C<next_key()> methods), you will get the I<target object's key>, not the
2365 ref's key. This gets even more interesting with the above example, where
2366 the I<circle> key points to the base DB object, which technically doesn't
2367 have a key. So I made DBM::Deep return "[base]" as the key name in that
2370 =head1 CAVEATS / ISSUES / BUGS
2372 This section describes all the known issues with DBM::Deep. It you have found
2373 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2375 =head2 UNUSED SPACE RECOVERY
2377 One major caveat with DBM::Deep is that space occupied by existing keys and
2378 values is not recovered when they are deleted. Meaning if you keep deleting
2379 and adding new keys, your file will continuously grow. I am working on this,
2380 but in the meantime you can call the built-in C<optimize()> method from time to
2381 time (perhaps in a crontab or something) to recover all your unused space.
2383 $db->optimize(); # returns true on success
2385 This rebuilds the ENTIRE database into a new file, then moves it on top of
2386 the original. The new file will have no unused space, thus it will take up as
2387 little disk space as possible. Please note that this operation can take
2388 a long time for large files, and you need enough disk space to temporarily hold
2389 2 copies of your DB file. The temporary file is created in the same directory
2390 as the original, named with a ".tmp" extension, and is deleted when the
2391 operation completes. Oh, and if locking is enabled, the DB is automatically
2392 locked for the entire duration of the copy.
2396 B<WARNING:> Only call optimize() on the top-level node of the database, and
2397 make sure there are no child references lying around. DBM::Deep keeps a reference
2398 counter, and if it is greater than 1, optimize() will abort and return undef.
2400 =head2 AUTOVIVIFICATION
2402 Unfortunately, autovivification doesn't work with tied hashes. This appears to
2403 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
2404 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
2405 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
2406 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
2409 $db->{foo}->{bar} = "hello";
2411 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
2412 being an empty hash. Try this instead, which works fine:
2414 $db->{foo} = { bar => "hello" };
2416 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
2417 the execution path, and Perl indeed passes an empty hash to the STORE() method.
2418 Probably a bug in Perl.
2420 =head2 FILE CORRUPTION
2422 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2423 for a 32-bit signature when opened, but other corruption in files can cause
2424 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2425 stuck in an infinite loop depending on the level of corruption. File write
2426 operations are not checked for failure (for speed), so if you happen to run
2427 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2428 be addressed in a later version of DBM::Deep.
2432 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2433 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2434 about setting up your NFS server with a locking daemon, then using lockf() to
2435 lock your files, but your milage may vary there as well. From what I
2436 understand, there is no real way to do it. However, if you need access to the
2437 underlying FileHandle in DBM::Deep for using some other kind of locking scheme like
2438 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2440 =head2 COPYING OBJECTS
2442 Beware of copying tied objects in Perl. Very strange things can happen.
2443 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2444 returns a new, blessed, tied hash or array to the same level in the DB.
2446 my $copy = $db->clone();
2450 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2451 These functions cause every element in the array to move, which can be murder
2452 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2453 a different location. This may be addressed in a later version.
2457 This section discusses DBM::Deep's speed and memory usage.
2461 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2462 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2463 multi-level hash/array support, and cross-platform FTPable files. Even so,
2464 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2465 with huge databases. Here is some test data:
2467 Adding 1,000,000 keys to new DB file...
2469 At 100 keys, avg. speed is 2,703 keys/sec
2470 At 200 keys, avg. speed is 2,642 keys/sec
2471 At 300 keys, avg. speed is 2,598 keys/sec
2472 At 400 keys, avg. speed is 2,578 keys/sec
2473 At 500 keys, avg. speed is 2,722 keys/sec
2474 At 600 keys, avg. speed is 2,628 keys/sec
2475 At 700 keys, avg. speed is 2,700 keys/sec
2476 At 800 keys, avg. speed is 2,607 keys/sec
2477 At 900 keys, avg. speed is 2,190 keys/sec
2478 At 1,000 keys, avg. speed is 2,570 keys/sec
2479 At 2,000 keys, avg. speed is 2,417 keys/sec
2480 At 3,000 keys, avg. speed is 1,982 keys/sec
2481 At 4,000 keys, avg. speed is 1,568 keys/sec
2482 At 5,000 keys, avg. speed is 1,533 keys/sec
2483 At 6,000 keys, avg. speed is 1,787 keys/sec
2484 At 7,000 keys, avg. speed is 1,977 keys/sec
2485 At 8,000 keys, avg. speed is 2,028 keys/sec
2486 At 9,000 keys, avg. speed is 2,077 keys/sec
2487 At 10,000 keys, avg. speed is 2,031 keys/sec
2488 At 20,000 keys, avg. speed is 1,970 keys/sec
2489 At 30,000 keys, avg. speed is 2,050 keys/sec
2490 At 40,000 keys, avg. speed is 2,073 keys/sec
2491 At 50,000 keys, avg. speed is 1,973 keys/sec
2492 At 60,000 keys, avg. speed is 1,914 keys/sec
2493 At 70,000 keys, avg. speed is 2,091 keys/sec
2494 At 80,000 keys, avg. speed is 2,103 keys/sec
2495 At 90,000 keys, avg. speed is 1,886 keys/sec
2496 At 100,000 keys, avg. speed is 1,970 keys/sec
2497 At 200,000 keys, avg. speed is 2,053 keys/sec
2498 At 300,000 keys, avg. speed is 1,697 keys/sec
2499 At 400,000 keys, avg. speed is 1,838 keys/sec
2500 At 500,000 keys, avg. speed is 1,941 keys/sec
2501 At 600,000 keys, avg. speed is 1,930 keys/sec
2502 At 700,000 keys, avg. speed is 1,735 keys/sec
2503 At 800,000 keys, avg. speed is 1,795 keys/sec
2504 At 900,000 keys, avg. speed is 1,221 keys/sec
2505 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2507 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2508 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2509 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2510 Run time was 12 min 3 sec.
2514 One of the great things about DBM::Deep is that it uses very little memory.
2515 Even with huge databases (1,000,000+ keys) you will not see much increased
2516 memory on your process. DBM::Deep relies solely on the filesystem for storing
2517 and fetching data. Here is output from I</usr/bin/top> before even opening a
2520 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2521 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2523 Basically the process is taking 2,716K of memory. And here is the same
2524 process after storing and fetching 1,000,000 keys:
2526 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2527 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2529 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2530 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2532 =head1 DB FILE FORMAT
2534 In case you were interested in the underlying DB file format, it is documented
2535 here in this section. You don't need to know this to use the module, it's just
2536 included for reference.
2540 DBM::Deep files always start with a 32-bit signature to identify the file type.
2541 This is at offset 0. The signature is "DPDB" in network byte order. This is
2542 checked when the file is opened.
2546 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2547 has a standard header containing the type of data, the length of data, and then
2548 the data itself. The type is a single character (1 byte), the length is a
2549 32-bit unsigned long in network byte order, and the data is, well, the data.
2550 Here is how it unfolds:
2554 Immediately after the 32-bit file signature is the I<Master Index> record.
2555 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2556 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2557 depending on how the DBM::Deep object was constructed.
2561 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2562 number). The first 8-bit char of the MD5 signature is the offset into the
2563 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2564 index element is a file offset of the next tag for the key/element in question,
2565 which is usually a I<Bucket List> tag (see below).
2569 The next tag I<could> be another index, depending on how many keys/elements
2570 exist. See L<RE-INDEXING> below for details.
2574 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2575 file offsets to where the actual data is stored. It starts with a standard
2576 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2577 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2578 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2579 When the list fills up, a I<Re-Index> operation is performed (See
2580 L<RE-INDEXING> below).
2584 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2585 index/value pair (in array mode). It starts with a standard tag header with
2586 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2587 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2588 header. The size reported in the tag header is only for the value, but then,
2589 just after the value is another size (32-bit unsigned long) and then the plain
2590 key itself. Since the value is likely to be fetched more often than the plain
2591 key, I figured it would be I<slightly> faster to store the value first.
2595 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2596 record for the nested structure, where the process begins all over again.
2600 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2601 exhausted. Then, when another key/element comes in, the list is converted to a
2602 new index record. However, this index will look at the next char in the MD5
2603 hash, and arrange new Bucket List pointers accordingly. This process is called
2604 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2605 17 (16 + new one) keys/elements are removed from the old Bucket List and
2606 inserted into the new index. Several new Bucket Lists are created in the
2607 process, as a new MD5 char from the key is being examined (it is unlikely that
2608 the keys will all share the same next char of their MD5s).
2612 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2613 when the Bucket Lists will turn into indexes, but the first round tends to
2614 happen right around 4,000 keys. You will see a I<slight> decrease in
2615 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2616 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2617 right around 900,000 keys. This process can continue nearly indefinitely --
2618 right up until the point the I<MD5> signatures start colliding with each other,
2619 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2620 getting struck by lightning while you are walking to cash in your tickets.
2621 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2622 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2623 this is 340 unodecillion, but don't quote me).
2627 When a new key/element is stored, the key (or index number) is first ran through
2628 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2629 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2630 for the first char of the signature (in this case I<b>). If it does not exist,
2631 a new I<Bucket List> is created for our key (and the next 15 future keys that
2632 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2633 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2634 this point, unless we are replacing an existing I<Bucket>), where the actual
2635 data will be stored.
2639 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2640 (or index number), then walking along the indexes. If there are enough
2641 keys/elements in this DB level, there might be nested indexes, each linked to
2642 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2643 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2644 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2645 plain key are stored.
2649 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2650 methods. In this process the indexes are walked systematically, and each key
2651 fetched in increasing MD5 order (which is why it appears random). Once the
2652 I<Bucket> is found, the value is skipped the plain key returned instead.
2653 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2654 alphabetically sorted. This only happens on an index-level -- as soon as the
2655 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2656 so it's pretty much undefined how the keys will come out -- just like Perl's
2659 =head1 CODE COVERAGE
2661 I use B<Devel::Cover> to test the code coverage of my tests, below is the B<Devel::Cover> report on this
2662 module's test suite.
2664 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2665 File stmt bran cond sub pod time total
2666 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2667 blib/lib/DBM/Deep.pm 94.1 82.9 74.5 98.0 10.5 98.1 88.2
2668 blib/lib/DBM/Deep/Array.pm 97.8 83.3 50.0 100.0 n/a 1.6 94.4
2669 blib/lib/DBM/Deep/Hash.pm 93.3 85.7 100.0 100.0 n/a 0.3 92.7
2670 Total 94.5 83.1 75.5 98.4 10.5 100.0 89.0
2671 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2675 Joseph Huckaby, L<jhuckaby@cpan.org>
2677 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2681 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2682 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2686 Copyright (c) 2002-2005 Joseph Huckaby. All Rights Reserved.
2687 This is free software, you may use it and distribute it under the
2688 same terms as Perl itself.