7 # Multi-level database module for storing hash trees, arrays and simple
8 # key/value pairs into FTP-able, cross-platform binary database files.
10 # Type `perldoc DBM::Deep` for complete documentation.
14 # tie %db, 'DBM::Deep', 'my_database.db'; # standard tie() method
16 # my $db = new DBM::Deep( 'my_database.db' ); # preferred OO method
18 # $db->{my_scalar} = 'hello world';
19 # $db->{my_hash} = { larry => 'genius', hashes => 'fast' };
20 # $db->{my_array} = [ 1, 2, 3, time() ];
21 # $db->{my_complex} = [ 'hello', { perl => 'rules' }, 42, 99 ];
22 # push @{$db->{my_array}}, 'another value';
23 # my @key_list = keys %{$db->{my_hash}};
24 # print "This module " . $db->{my_complex}->[1]->{perl} . "!\n";
27 # (c) 2002-2005 Joseph Huckaby. All Rights Reserved.
28 # This program is free software; you can redistribute it and/or
29 # modify it under the same terms as Perl itself.
34 use Fcntl qw( :DEFAULT :flock :seek );
38 use vars qw( $VERSION );
42 # Set to 4 and 'N' for 32-bit offset tags (default). Theoretical limit of 4 GB per file.
43 # (Perl must be compiled with largefile support for files > 2 GB)
45 # Set to 8 and 'Q' for 64-bit offsets. Theoretical limit of 16 XB per file.
46 # (Perl must be compiled with largefile and 64-bit long support)
52 # Set to 4 and 'N' for 32-bit data length prefixes. Limit of 4 GB for each key/value.
53 # Upgrading this is possible (see above) but probably not necessary. If you need
54 # more than 4 GB for a single key or value, this module is really not for you :-)
56 #my $DATA_LENGTH_SIZE = 4;
57 #my $DATA_LENGTH_PACK = 'N';
58 our ($LONG_SIZE, $LONG_PACK, $DATA_LENGTH_SIZE, $DATA_LENGTH_PACK);
61 # Maximum number of buckets per list before another level of indexing is done.
62 # Increase this value for slightly greater speed, but larger database files.
63 # DO NOT decrease this value below 16, due to risk of recursive reindex overrun.
68 # Better not adjust anything below here, unless you're me :-)
72 # Setup digest function for keys
74 our ($DIGEST_FUNC, $HASH_SIZE);
75 #my $DIGEST_FUNC = \&Digest::MD5::md5;
78 # Precalculate index and bucket sizes based on values above.
81 my ($INDEX_SIZE, $BUCKET_SIZE, $BUCKET_LIST_SIZE);
88 # Setup file and tag signatures. These should never change.
90 sub SIG_FILE () { 'DPDB' }
91 sub SIG_HASH () { 'H' }
92 sub SIG_ARRAY () { 'A' }
93 sub SIG_SCALAR () { 'S' }
94 sub SIG_NULL () { 'N' }
95 sub SIG_DATA () { 'D' }
96 sub SIG_INDEX () { 'I' }
97 sub SIG_BLIST () { 'B' }
101 # Setup constants for users to pass to new()
103 sub TYPE_HASH () { return SIG_HASH; }
104 sub TYPE_ARRAY () { return SIG_ARRAY; }
105 sub TYPE_SCALAR () { return SIG_SCALAR; }
109 # Class constructor method for Perl OO interface.
110 # Calls tie() and returns blessed reference to tied hash or array,
111 # providing a hybrid OO/tie interface.
115 if (scalar(@_) > 1) { $args = {@_}; }
116 else { $args = { file => shift }; }
119 # Check if we want a tied hash or array.
122 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
123 $class = 'DBM::Deep::Array';
124 require DBM::Deep::Array;
125 tie @$self, $class, %$args;
128 $class = 'DBM::Deep::Hash';
129 require DBM::Deep::Hash;
130 tie %$self, $class, %$args;
133 return bless $self, $class;
138 # Setup $self and bless into this class.
143 # These are the defaults to be optionally overridden below
146 base_offset => length(SIG_FILE),
149 foreach my $param ( keys %$self ) {
150 next unless exists $args->{$param};
151 $self->{$param} = delete $args->{$param}
154 $self->{root} = exists $args->{root}
156 : DBM::Deep::_::Root->new( $args );
158 if (!defined($self->fh)) { $self->_open(); }
165 require DBM::Deep::Hash;
166 return DBM::Deep::Hash->TIEHASH( @_ );
171 require DBM::Deep::Array;
172 return DBM::Deep::Array->TIEARRAY( @_ );
175 #XXX Unneeded now ...
181 # Open a FileHandle to the database, create if nonexistent.
182 # Make sure file signature matches DeepDB spec.
184 my $self = $_[0]->_get_self;
186 if (defined($self->fh)) { $self->_close(); }
189 # Theoretically, adding O_BINARY should remove the need for the binmode
190 # Of course, testing it is going to be ... interesting.
191 my $flags = O_RDWR | O_CREAT | O_BINARY;
193 #XXX Can the mode be anything but r+, w+, or a+??
194 #XXX ie, it has to be in read-write mode
195 #XXX So, should we verify that the mode is legitimate?
197 #XXX Maybe the mode thingy should just go away. There's no good
198 #XXX reason for it ...
199 if ( $self->root->{mode} eq 'w+' ) {
204 sysopen( $fh, $self->root->{file}, $flags )
206 $self->root->{fh} = $fh;
207 }; if ($@ ) { $self->_throw_error( "Received error: $@\n" ); }
208 if (! defined($self->fh)) {
209 return $self->_throw_error("Cannot sysopen file: " . $self->root->{file} . ": $!");
214 #XXX Can we remove this by using the right sysopen() flags?
215 # Maybe ... q.v. above
216 binmode $fh; # for win32
218 if ($self->root->{autoflush}) {
219 my $old = select $fh;
225 seek($fh, 0, SEEK_SET);
228 my $bytes_read = read( $fh, $signature, length(SIG_FILE));
231 # File is empty -- write signature and master index
234 seek($fh, 0, SEEK_SET);
236 $self->root->{end} = length(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 );
241 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
243 # Flush the filehandle
244 my $old_fh = select $fh;
254 # Check signature was valid
256 unless ($signature eq SIG_FILE) {
258 return $self->_throw_error("Signature not found -- file is not a Deep DB");
261 $self->root->{end} = (stat($fh))[7];
264 # Get our type from master index signature
266 my $tag = $self->_load_tag($self->base_offset);
268 #XXX We probably also want to store the hash algorithm name and not assume anything
269 #XXX The cool thing would be to allow a different hashing algorithm at every level
272 return $self->_throw_error("Corrupted file, no master index record");
274 if ($self->{type} ne $tag->{signature}) {
275 return $self->_throw_error("File type mismatch");
283 # Close database FileHandle
285 my $self = $_[0]->_get_self;
286 close $self->root->{fh};
291 # Given offset, signature and content, create tag and write to disk
293 my ($self, $offset, $sig, $content) = @_;
294 my $size = length($content);
298 seek($fh, $offset, SEEK_SET);
299 print($fh $sig . pack($DATA_LENGTH_PACK, $size) . $content );
301 if ($offset == $self->root->{end}) {
302 $self->root->{end} += SIG_SIZE + $DATA_LENGTH_SIZE + $size;
308 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
315 # Given offset, load single tag and return signature, size and data
322 seek($fh, $offset, SEEK_SET);
323 if (eof $fh) { return undef; }
326 read( $fh, $sig, SIG_SIZE);
329 read( $fh, $size, $DATA_LENGTH_SIZE);
330 $size = unpack($DATA_LENGTH_PACK, $size);
333 read( $fh, $buffer, $size);
338 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
345 # Given index tag, lookup single entry in index and return .
348 my ($tag, $index) = @_;
350 my $location = unpack($LONG_PACK, substr($tag->{content}, $index * $LONG_SIZE, $LONG_SIZE) );
351 if (!$location) { return; }
353 return $self->_load_tag( $location );
358 # Adds one key/value pair to bucket list, given offset, MD5 digest of key,
359 # plain (undigested) key and value.
362 my ($tag, $md5, $plain_key, $value) = @_;
363 my $keys = $tag->{content};
367 my $is_dbm_deep = eval { $value->isa( 'DBM::Deep' ) };
368 my $internal_ref = $is_dbm_deep && ($value->root eq $self->root);
373 # Iterate through buckets, seeing if this is a new entry or a replace.
375 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
376 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
377 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
380 # Found empty bucket (end of list). Populate and exit loop.
384 $location = $internal_ref
385 ? $value->base_offset
386 : $self->root->{end};
388 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
389 print($fh $md5 . pack($LONG_PACK, $location) );
392 elsif ($md5 eq $key) {
394 # Found existing bucket with same key. Replace with new value.
399 $location = $value->base_offset;
400 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
401 print($fh $md5 . pack($LONG_PACK, $location) );
404 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
406 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
409 # If value is a hash, array, or raw value with equal or less size, we can
410 # reuse the same content area of the database. Otherwise, we have to create
411 # a new content area at the EOF.
414 my $r = Scalar::Util::reftype( $value ) || '';
415 if ( $r eq 'HASH' || $r eq 'ARRAY' ) { $actual_length = $INDEX_SIZE; }
416 else { $actual_length = length($value); }
418 if ($actual_length <= $size) {
422 $location = $self->root->{end};
423 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE, SEEK_SET);
424 print($fh pack($LONG_PACK, $location) );
432 # If this is an internal reference, return now.
433 # No need to write value or plain key
440 # If bucket didn't fit into list, split into a new index level
443 seek($fh, $tag->{ref_loc}, SEEK_SET);
444 print($fh pack($LONG_PACK, $self->root->{end}) );
446 my $index_tag = $self->_create_tag($self->root->{end}, SIG_INDEX, chr(0) x $INDEX_SIZE);
449 $keys .= $md5 . pack($LONG_PACK, 0);
451 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
452 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
454 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
455 my $num = ord(substr($key, $tag->{ch} + 1, 1));
457 if ($offsets[$num]) {
458 my $offset = $offsets[$num] + SIG_SIZE + $DATA_LENGTH_SIZE;
459 seek($fh, $offset, SEEK_SET);
461 read( $fh, $subkeys, $BUCKET_LIST_SIZE);
463 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
464 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
466 seek($fh, $offset + ($k * $BUCKET_SIZE), SEEK_SET);
467 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
473 $offsets[$num] = $self->root->{end};
474 seek($fh, $index_tag->{offset} + ($num * $LONG_SIZE), SEEK_SET);
475 print($fh pack($LONG_PACK, $self->root->{end}) );
477 my $blist_tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
479 seek($fh, $blist_tag->{offset}, SEEK_SET);
480 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
485 $location ||= $self->root->{end};
486 } # re-index bucket list
489 # Seek to content area and store signature, value and plaintext key
493 seek($fh, $location, SEEK_SET);
496 # Write signature based on content type, set content length and write actual value.
498 my $r = Scalar::Util::reftype($value) || '';
500 print($fh TYPE_HASH );
501 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
502 $content_length = $INDEX_SIZE;
504 elsif ($r eq 'ARRAY') {
505 print($fh TYPE_ARRAY );
506 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
507 $content_length = $INDEX_SIZE;
509 elsif (!defined($value)) {
510 print($fh SIG_NULL );
511 print($fh pack($DATA_LENGTH_PACK, 0) );
515 print($fh SIG_DATA );
516 print($fh pack($DATA_LENGTH_PACK, length($value)) . $value );
517 $content_length = length($value);
521 # Plain key is stored AFTER value, as keys are typically fetched less often.
523 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
526 # If value is blessed, preserve class name
528 if ( $self->root->{autobless} ) {
529 my $value_class = Scalar::Util::blessed($value);
530 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
532 # Blessed ref -- will restore later
535 print($fh pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
536 $content_length += 1;
537 $content_length += $DATA_LENGTH_SIZE + length($value_class);
541 $content_length += 1;
546 # If this is a new content area, advance EOF counter
548 if ($location == $self->root->{end}) {
549 $self->root->{end} += SIG_SIZE;
550 $self->root->{end} += $DATA_LENGTH_SIZE + $content_length;
551 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
555 # If content is a hash or array, create new child DeepDB object and
556 # pass each key or element to it.
559 my $branch = DBM::Deep->new(
561 base_offset => $location,
564 foreach my $key (keys %{$value}) {
565 #$branch->{$key} = $value->{$key};
566 $branch->STORE( $key, $value->{$key} );
569 elsif ($r eq 'ARRAY') {
570 my $branch = DBM::Deep->new(
572 base_offset => $location,
576 foreach my $element (@{$value}) {
577 #$branch->[$index] = $element;
578 $branch->STORE( $index, $element );
586 return $self->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
589 sub _get_bucket_value {
591 # Fetch single value given tag and MD5 digested key.
594 my ($tag, $md5) = @_;
595 my $keys = $tag->{content};
600 # Iterate through buckets, looking for a key match
603 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
604 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
605 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
609 # Hit end of list, no match
614 if ( $md5 ne $key ) {
619 # Found match -- seek to offset and read signature
622 seek($fh, $subloc, SEEK_SET);
623 read( $fh, $signature, SIG_SIZE);
626 # If value is a hash or array, return new DeepDB object with correct offset
628 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
629 my $obj = DBM::Deep->new(
631 base_offset => $subloc,
635 if ($self->root->{autobless}) {
637 # Skip over value and plain key to see if object needs
640 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
643 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
644 if ($size) { seek($fh, $size, SEEK_CUR); }
647 read( $fh, $bless_bit, 1);
648 if (ord($bless_bit)) {
650 # Yes, object needs to be re-blessed
653 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
654 if ($size) { read( $fh, $class_name, $size); }
655 if ($class_name) { $obj = bless( $obj, $class_name ); }
663 # Otherwise return actual value
665 elsif ($signature eq SIG_DATA) {
668 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
669 if ($size) { read( $fh, $value, $size); }
674 # Key exists, but content is null
684 # Delete single key/value pair given tag and MD5 digested key.
687 my ($tag, $md5) = @_;
688 my $keys = $tag->{content};
693 # Iterate through buckets, looking for a key match
696 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
697 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
698 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
702 # Hit end of list, no match
707 if ( $md5 ne $key ) {
712 # Matched key -- delete bucket and return
714 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
715 print($fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
716 print($fh chr(0) x $BUCKET_SIZE );
726 # Check existence of single key given tag and MD5 digested key.
729 my ($tag, $md5) = @_;
730 my $keys = $tag->{content};
733 # Iterate through buckets, looking for a key match
736 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
737 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
738 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
742 # Hit end of list, no match
747 if ( $md5 ne $key ) {
752 # Matched key -- return true
760 sub _find_bucket_list {
762 # Locate offset for bucket list, given digested key
768 # Locate offset for bucket list using digest index system
771 my $tag = $self->_load_tag($self->base_offset);
772 if (!$tag) { return; }
774 while ($tag->{signature} ne SIG_BLIST) {
775 $tag = $self->_index_lookup($tag, ord(substr($md5, $ch, 1)));
776 if (!$tag) { return; }
783 sub _traverse_index {
785 # Scan index and recursively step into deeper levels, looking for next key.
787 my ($self, $offset, $ch, $force_return_next) = @_;
788 $force_return_next = undef unless $force_return_next;
790 my $tag = $self->_load_tag( $offset );
794 if ($tag->{signature} ne SIG_BLIST) {
795 my $content = $tag->{content};
797 if ($self->{return_next}) { $start = 0; }
798 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
800 for (my $index = $start; $index < 256; $index++) {
801 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
803 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
804 if (defined($result)) { return $result; }
808 $self->{return_next} = 1;
811 elsif ($tag->{signature} eq SIG_BLIST) {
812 my $keys = $tag->{content};
813 if ($force_return_next) { $self->{return_next} = 1; }
816 # Iterate through buckets, looking for a key match
818 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
819 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
820 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
824 # End of bucket list -- return to outer loop
826 $self->{return_next} = 1;
829 elsif ($key eq $self->{prev_md5}) {
831 # Located previous key -- return next one found
833 $self->{return_next} = 1;
836 elsif ($self->{return_next}) {
838 # Seek to bucket location and skip over signature
840 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
843 # Skip over value to get to plain key
846 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
847 if ($size) { seek($fh, $size, SEEK_CUR); }
850 # Read in plain key and return as scalar
853 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
854 if ($size) { read( $fh, $plain_key, $size); }
860 $self->{return_next} = 1;
861 } # tag is a bucket list
868 # Locate next key, given digested previous one
870 my $self = $_[0]->_get_self;
872 $self->{prev_md5} = $_[1] ? $_[1] : undef;
873 $self->{return_next} = 0;
876 # If the previous key was not specifed, start at the top and
877 # return the first one found.
879 if (!$self->{prev_md5}) {
880 $self->{prev_md5} = chr(0) x $HASH_SIZE;
881 $self->{return_next} = 1;
884 return $self->_traverse_index( $self->base_offset, 0 );
889 # If db locking is set, flock() the db file. If called multiple
890 # times before unlock(), then the same number of unlocks() must
891 # be called before the lock is released.
893 my $self = $_[0]->_get_self;
895 $type = LOCK_EX unless defined $type;
897 if ($self->root->{locking}) {
898 if (!$self->root->{locked}) { flock($self->fh, $type); }
899 $self->root->{locked}++;
909 # If db locking is set, unlock the db file. See note in lock()
910 # regarding calling lock() multiple times.
912 my $self = $_[0]->_get_self;
914 if ($self->root->{locking} && $self->root->{locked} > 0) {
915 $self->root->{locked}--;
916 if (!$self->root->{locked}) { flock($self->fh, LOCK_UN); }
924 #XXX These uses of ref() need verified
927 # Copy single level of keys or elements to new DB handle.
928 # Recurse for nested structures
930 my $self = $_[0]->_get_self;
933 if ($self->type eq TYPE_HASH) {
934 my $key = $self->first_key();
936 my $value = $self->get($key);
937 #XXX This doesn't work with autobless
938 if (!ref($value)) { $db_temp->{$key} = $value; }
940 my $type = $value->type;
941 if ($type eq TYPE_HASH) { $db_temp->{$key} = {}; }
942 else { $db_temp->{$key} = []; }
943 $value->_copy_node( $db_temp->{$key} );
945 $key = $self->next_key($key);
949 my $length = $self->length();
950 for (my $index = 0; $index < $length; $index++) {
951 my $value = $self->get($index);
952 if (!ref($value)) { $db_temp->[$index] = $value; }
953 #XXX NO tests for this code
955 my $type = $value->type;
956 if ($type eq TYPE_HASH) { $db_temp->[$index] = {}; }
957 else { $db_temp->[$index] = []; }
958 $value->_copy_node( $db_temp->[$index] );
966 # Recursively export into standard Perl hashes and arrays.
968 my $self = $_[0]->_get_self;
971 if ($self->type eq TYPE_HASH) { $temp = {}; }
972 elsif ($self->type eq TYPE_ARRAY) { $temp = []; }
975 $self->_copy_node( $temp );
983 # Recursively import Perl hash/array structure
985 #XXX This use of ref() seems to be ok
986 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
988 my $self = $_[0]->_get_self;
991 #XXX This use of ref() seems to be ok
994 # struct is not a reference, so just import based on our type
998 if ($self->type eq TYPE_HASH) { $struct = {@_}; }
999 elsif ($self->type eq TYPE_ARRAY) { $struct = [@_]; }
1002 my $r = Scalar::Util::reftype($struct) || '';
1003 if ($r eq "HASH" && $self->type eq TYPE_HASH) {
1004 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
1006 elsif ($r eq "ARRAY" && $self->type eq TYPE_ARRAY) {
1007 $self->push( @$struct );
1010 return $self->_throw_error("Cannot import: type mismatch");
1018 # Rebuild entire database into new file, then move
1019 # it back on top of original.
1021 my $self = $_[0]->_get_self;
1023 #XXX Need to create a new test for this
1024 # if ($self->root->{links} > 1) {
1025 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
1028 my $db_temp = DBM::Deep->new(
1029 file => $self->root->{file} . '.tmp',
1033 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
1037 $self->_copy_node( $db_temp );
1041 # Attempt to copy user, group and permissions over to new file
1043 my @stats = stat($self->fh);
1044 my $perms = $stats[2] & 07777;
1045 my $uid = $stats[4];
1046 my $gid = $stats[5];
1047 chown( $uid, $gid, $self->root->{file} . '.tmp' );
1048 chmod( $perms, $self->root->{file} . '.tmp' );
1050 # q.v. perlport for more information on this variable
1051 if ( $^O eq 'MSWin32' ) {
1053 # Potential race condition when optmizing on Win32 with locking.
1054 # The Windows filesystem requires that the filehandle be closed
1055 # before it is overwritten with rename(). This could be redone
1062 if (!rename $self->root->{file} . '.tmp', $self->root->{file}) {
1063 unlink $self->root->{file} . '.tmp';
1065 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
1077 # Make copy of object and return
1079 my $self = $_[0]->_get_self;
1081 return DBM::Deep->new(
1082 type => $self->type,
1083 base_offset => $self->base_offset,
1089 my %is_legal_filter = map {
1092 store_key store_value
1093 fetch_key fetch_value
1098 # Setup filter function for storing or fetching the key or value
1100 my $self = $_[0]->_get_self;
1101 my $type = lc $_[1];
1102 my $func = $_[2] ? $_[2] : undef;
1104 if ( $is_legal_filter{$type} ) {
1105 $self->root->{"filter_$type"} = $func;
1119 # Get access to the root structure
1121 my $self = $_[0]->_get_self;
1122 return $self->{root};
1127 # Get access to the raw FileHandle
1129 #XXX It will be useful, though, when we split out HASH and ARRAY
1130 my $self = $_[0]->_get_self;
1131 return $self->root->{fh};
1136 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
1138 my $self = $_[0]->_get_self;
1139 return $self->{type};
1144 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
1146 my $self = $_[0]->_get_self;
1147 return $self->{base_offset};
1152 # Get last error string, or undef if no error
1155 #? ( _get_self($_[0])->{root}->{error} or undef )
1156 ? ( $_[0]->_get_self->{root}->{error} or undef )
1166 # Store error string in self
1168 my $self = $_[0]->_get_self;
1169 my $error_text = $_[1];
1171 if ( Scalar::Util::blessed $self ) {
1172 $self->root->{error} = $error_text;
1174 unless ($self->root->{debug}) {
1175 die "DBM::Deep: $error_text\n";
1178 warn "DBM::Deep: $error_text\n";
1182 die "DBM::Deep: $error_text\n";
1190 my $self = $_[0]->_get_self;
1192 undef $self->root->{error};
1197 # Precalculate index, bucket and bucket list sizes
1200 #XXX I don't like this ...
1201 set_pack() unless defined $LONG_SIZE;
1203 $INDEX_SIZE = 256 * $LONG_SIZE;
1204 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
1205 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
1210 # Set pack/unpack modes (see file header for more)
1212 my ($long_s, $long_p, $data_s, $data_p) = @_;
1214 $LONG_SIZE = $long_s ? $long_s : 4;
1215 $LONG_PACK = $long_p ? $long_p : 'N';
1217 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
1218 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
1225 # Set key digest function (default is MD5)
1227 my ($digest_func, $hash_size) = @_;
1229 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
1230 $HASH_SIZE = $hash_size ? $hash_size : 16;
1236 # tie() methods (hashes and arrays)
1241 # Store single hash key/value or array element in database.
1243 my $self = $_[0]->_get_self;
1246 #XXX What is ref() checking here?
1247 #YYY User may be storing a hash, in which case we do not want it run
1248 #YYY through the filtering system
1249 my $value = ($self->root->{filter_store_value} && !ref($_[2]))
1250 ? $self->root->{filter_store_value}->($_[2])
1253 my $unpacked_key = $key;
1254 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1256 my $md5 = $DIGEST_FUNC->($key);
1259 # Make sure file is open
1261 if (!defined($self->fh) && !$self->_open()) {
1269 # Request exclusive lock for writing
1271 $self->lock( LOCK_EX );
1274 # If locking is enabled, set 'end' parameter again, in case another
1275 # DB instance appended to our file while we were unlocked.
1277 if ($self->root->{locking} || $self->root->{volatile}) {
1278 $self->root->{end} = (stat($fh))[7];
1282 # Locate offset for bucket list using digest index system
1284 my $tag = $self->_load_tag($self->base_offset);
1286 $tag = $self->_create_tag($self->base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1290 while ($tag->{signature} ne SIG_BLIST) {
1291 my $num = ord(substr($md5, $ch, 1));
1292 my $new_tag = $self->_index_lookup($tag, $num);
1294 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1295 seek($fh, $ref_loc, SEEK_SET);
1296 print($fh pack($LONG_PACK, $self->root->{end}) );
1298 $tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1299 $tag->{ref_loc} = $ref_loc;
1304 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1306 $tag->{ref_loc} = $ref_loc;
1313 # Add key/value to bucket list
1315 my $result = $self->_add_bucket( $tag, $md5, $key, $value );
1318 # If this object is an array, and bucket was not a replace, and key is numerical,
1319 # and index is equal or greater than current length, advance length variable.
1321 if (($result == 2) && ($self->type eq TYPE_ARRAY) && ($unpacked_key =~ /^\d+$/) && ($unpacked_key >= $self->FETCHSIZE())) {
1322 $self->STORESIZE( $unpacked_key + 1 );
1332 # Fetch single value or element given plain key or array index
1334 my $self = $_[0]->_get_self;
1337 if ( $self->type eq TYPE_HASH ) {
1338 if ( my $filter = $self->root->{filter_store_key} ) {
1339 $key = $filter->( $key );
1342 elsif ( $self->type eq TYPE_ARRAY ) {
1343 if ( $key =~ /^\d+$/ ) {
1344 $key = pack($LONG_PACK, $key);
1348 my $md5 = $DIGEST_FUNC->($key);
1351 # Make sure file is open
1353 if (!defined($self->fh)) { $self->_open(); }
1356 # Request shared lock for reading
1358 $self->lock( LOCK_SH );
1360 my $tag = $self->_find_bucket_list( $md5 );
1367 # Get value from bucket list
1369 my $result = $self->_get_bucket_value( $tag, $md5 );
1373 #XXX What is ref() checking here?
1374 return ($result && !ref($result) && $self->root->{filter_fetch_value}) ? $self->root->{filter_fetch_value}->($result) : $result;
1379 # Delete single key/value pair or element given plain key or array index
1381 my $self = $_[0]->_get_self;
1382 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1384 my $unpacked_key = $key;
1385 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1386 my $md5 = $DIGEST_FUNC->($key);
1389 # Make sure file is open
1391 if (!defined($self->fh)) { $self->_open(); }
1394 # Request exclusive lock for writing
1396 $self->lock( LOCK_EX );
1398 my $tag = $self->_find_bucket_list( $md5 );
1407 my $value = $self->FETCH( $unpacked_key );
1408 my $result = $self->_delete_bucket( $tag, $md5 );
1411 # If this object is an array and the key deleted was on the end of the stack,
1412 # decrement the length variable.
1414 if ($result && ($self->type eq TYPE_ARRAY) && ($unpacked_key == $self->FETCHSIZE() - 1)) {
1415 $self->STORESIZE( $unpacked_key );
1425 # Check if a single key or element exists given plain key or array index
1427 my $self = $_[0]->_get_self;
1428 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1430 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1431 my $md5 = $DIGEST_FUNC->($key);
1434 # Make sure file is open
1436 if (!defined($self->fh)) { $self->_open(); }
1439 # Request shared lock for reading
1441 $self->lock( LOCK_SH );
1443 my $tag = $self->_find_bucket_list( $md5 );
1446 # For some reason, the built-in exists() function returns '' for false
1454 # Check if bucket exists and return 1 or ''
1456 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1465 # Clear all keys from hash, or all elements from array.
1467 my $self = $_[0]->_get_self;
1470 # Make sure file is open
1472 if (!defined($self->fh)) { $self->_open(); }
1475 # Request exclusive lock for writing
1477 $self->lock( LOCK_EX );
1481 seek($fh, $self->base_offset, SEEK_SET);
1487 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
1495 # Public method aliases
1497 *put = *store = *STORE;
1498 *get = *fetch = *FETCH;
1503 package DBM::Deep::_::Root;
1518 filter_store_key => undef,
1519 filter_store_value => undef,
1520 filter_fetch_key => undef,
1521 filter_fetch_value => undef,
1532 return unless $self;
1534 close $self->{fh} if $self->{fh};
1545 DBM::Deep - A pure perl multi-level hash/array DBM
1550 my $db = DBM::Deep->new( "foo.db" );
1552 $db->{key} = 'value'; # tie() style
1555 $db->put('key', 'value'); # OO style
1556 print $db->get('key');
1558 # true multi-level support
1559 $db->{my_complex} = [
1560 'hello', { perl => 'rules' },
1565 A unique flat-file database module, written in pure perl. True
1566 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1567 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1568 handle millions of keys and unlimited hash levels without significant
1569 slow-down. Written from the ground-up in pure perl -- this is NOT a
1570 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1571 Mac OS X and Windows.
1575 Hopefully you are using CPAN's excellent Perl module, which will download
1576 and install the module for you. If not, get the tarball, and run these
1588 Construction can be done OO-style (which is the recommended way), or using
1589 Perl's tie() function. Both are examined here.
1591 =head2 OO CONSTRUCTION
1593 The recommended way to construct a DBM::Deep object is to use the new()
1594 method, which gets you a blessed, tied hash or array reference.
1596 my $db = DBM::Deep->new( "foo.db" );
1598 This opens a new database handle, mapped to the file "foo.db". If this
1599 file does not exist, it will automatically be created. DB files are
1600 opened in "r+" (read/write) mode, and the type of object returned is a
1601 hash, unless otherwise specified (see L<OPTIONS> below).
1605 You can pass a number of options to the constructor to specify things like
1606 locking, autoflush, etc. This is done by passing an inline hash:
1608 my $db = DBM::Deep->new(
1614 Notice that the filename is now specified I<inside> the hash with
1615 the "file" parameter, as opposed to being the sole argument to the
1616 constructor. This is required if any options are specified.
1617 See L<OPTIONS> below for the complete list.
1621 You can also start with an array instead of a hash. For this, you must
1622 specify the C<type> parameter:
1624 my $db = DBM::Deep->new(
1626 type => DBM::Deep->TYPE_ARRAY
1629 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1630 a new DB file. If you create a DBM::Deep object with an existing file, the
1631 C<type> will be loaded from the file header, and ignored if it is passed
1634 =head2 TIE CONSTRUCTION
1636 Alternatively, you can create a DBM::Deep handle by using Perl's built-in
1637 tie() function. This is not ideal, because you get only a basic, tied hash
1638 (or array) which is not blessed, so you can't call any functions on it.
1641 tie %hash, "DBM::Deep", "foo.db";
1644 tie @array, "DBM::Deep", "bar.db";
1646 As with the OO constructor, you can replace the DB filename parameter with
1647 a hash containing one or more options (see L<OPTIONS> just below for the
1650 tie %hash, "DBM::Deep", {
1658 There are a number of options that can be passed in when constructing your
1659 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1665 Filename of the DB file to link the handle to. You can pass a full absolute
1666 filesystem path, partial path, or a plain filename if the file is in the
1667 current working directory. This is a required parameter.
1671 File open mode (read-only, read-write, etc.) string passed to Perl's FileHandle
1672 module. This is an optional parameter, and defaults to "r+" (read/write).
1673 B<Note:> If the default (r+) mode is selected, the file will also be auto-
1674 created if it doesn't exist.
1678 This parameter specifies what type of object to create, a hash or array. Use
1679 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1680 This only takes effect when beginning a new file. This is an optional
1681 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1685 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1686 function to lock the database in exclusive mode for writes, and shared mode for
1687 reads. Pass any true value to enable. This affects the base DB handle I<and
1688 any child hashes or arrays> that use the same DB file. This is an optional
1689 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1693 Specifies whether autoflush is to be enabled on the underlying FileHandle.
1694 This obviously slows down write operations, but is required if you may have
1695 multiple processes accessing the same DB file (also consider enable I<locking>
1696 or at least I<volatile>). Pass any true value to enable. This is an optional
1697 parameter, and defaults to 0 (disabled).
1701 If I<volatile> mode is enabled, DBM::Deep will stat() the DB file before each
1702 STORE() operation. This is required if an outside force may change the size of
1703 the file between transactions. Locking also implicitly enables volatile. This
1704 is useful if you want to use a different locking system or write your own. Pass
1705 any true value to enable. This is an optional parameter, and defaults to 0
1710 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1711 restore them when fetched. This is an B<experimental> feature, and does have
1712 side-effects. Basically, when hashes are re-blessed into their original
1713 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1714 able to call any DBM::Deep methods on them. You have been warned.
1715 This is an optional parameter, and defaults to 0 (disabled).
1719 See L<FILTERS> below.
1723 Setting I<debug> mode will make all errors non-fatal, dump them out to
1724 STDERR, and continue on. This is for debugging purposes only, and probably
1725 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1729 Instead of passing a file path, you can instead pass a handle to an pre-opened
1730 filehandle. Note: Beware of using the magick *DATA handle, as this actually
1731 contains your entire Perl script, as well as the data following the __DATA__
1732 marker. This will not work, because DBM::Deep uses absolute seek()s into the
1733 file. Instead, consider reading *DATA into an IO::Scalar handle, then passing
1738 =head1 TIE INTERFACE
1740 With DBM::Deep you can access your databases using Perl's standard hash/array
1741 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can treat
1742 them as such. DBM::Deep will intercept all reads/writes and direct them to the right
1743 place -- the DB file. This has nothing to do with the L<TIE CONSTRUCTION>
1744 section above. This simply tells you how to use DBM::Deep using regular hashes
1745 and arrays, rather than calling functions like C<get()> and C<put()> (although those
1746 work too). It is entirely up to you how to want to access your databases.
1750 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1751 or even nested hashes (or arrays) using standard Perl syntax:
1753 my $db = DBM::Deep->new( "foo.db" );
1755 $db->{mykey} = "myvalue";
1757 $db->{myhash}->{subkey} = "subvalue";
1759 print $db->{myhash}->{subkey} . "\n";
1761 You can even step through hash keys using the normal Perl C<keys()> function:
1763 foreach my $key (keys %$db) {
1764 print "$key: " . $db->{$key} . "\n";
1767 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1768 pushes them onto an array, all before the loop even begins. If you have an
1769 extra large hash, this may exhaust Perl's memory. Instead, consider using
1770 Perl's C<each()> function, which pulls keys/values one at a time, using very
1773 while (my ($key, $value) = each %$db) {
1774 print "$key: $value\n";
1777 Please note that when using C<each()>, you should always pass a direct
1778 hash reference, not a lookup. Meaning, you should B<never> do this:
1781 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1783 This causes an infinite loop, because for each iteration, Perl is calling
1784 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1785 it effectively keeps returning the first key over and over again. Instead,
1786 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1790 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1791 reference. This includes inserting, removing and manipulating elements,
1792 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1793 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1794 or simply be a nested array reference inside a hash. Example:
1796 my $db = DBM::Deep->new(
1797 file => "foo-array.db",
1798 type => DBM::Deep->TYPE_ARRAY
1802 push @$db, "bar", "baz";
1803 unshift @$db, "bah";
1805 my $last_elem = pop @$db; # baz
1806 my $first_elem = shift @$db; # bah
1807 my $second_elem = $db->[1]; # bar
1809 my $num_elements = scalar @$db;
1813 In addition to the I<tie()> interface, you can also use a standard OO interface
1814 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1815 array) has its own methods, but both types share the following common methods:
1816 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1822 Stores a new hash key/value pair, or sets an array element value. Takes two
1823 arguments, the hash key or array index, and the new value. The value can be
1824 a scalar, hash ref or array ref. Returns true on success, false on failure.
1826 $db->put("foo", "bar"); # for hashes
1827 $db->put(1, "bar"); # for arrays
1831 Fetches the value of a hash key or array element. Takes one argument: the hash
1832 key or array index. Returns a scalar, hash ref or array ref, depending on the
1835 my $value = $db->get("foo"); # for hashes
1836 my $value = $db->get(1); # for arrays
1840 Checks if a hash key or array index exists. Takes one argument: the hash key
1841 or array index. Returns true if it exists, false if not.
1843 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1844 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1848 Deletes one hash key/value pair or array element. Takes one argument: the hash
1849 key or array index. Returns true on success, false if not found. For arrays,
1850 the remaining elements located after the deleted element are NOT moved over.
1851 The deleted element is essentially just undefined, which is exactly how Perl's
1852 internal arrays work. Please note that the space occupied by the deleted
1853 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1854 below for details and workarounds.
1856 $db->delete("foo"); # for hashes
1857 $db->delete(1); # for arrays
1861 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1862 value. Please note that the space occupied by the deleted keys/values or
1863 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1864 details and workarounds.
1866 $db->clear(); # hashes or arrays
1872 For hashes, DBM::Deep supports all the common methods described above, and the
1873 following additional methods: C<first_key()> and C<next_key()>.
1879 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1880 fetched in an undefined order (which appears random). Takes no arguments,
1881 returns the key as a scalar value.
1883 my $key = $db->first_key();
1887 Returns the "next" key in the hash, given the previous one as the sole argument.
1888 Returns undef if there are no more keys to be fetched.
1890 $key = $db->next_key($key);
1894 Here are some examples of using hashes:
1896 my $db = DBM::Deep->new( "foo.db" );
1898 $db->put("foo", "bar");
1899 print "foo: " . $db->get("foo") . "\n";
1901 $db->put("baz", {}); # new child hash ref
1902 $db->get("baz")->put("buz", "biz");
1903 print "buz: " . $db->get("baz")->get("buz") . "\n";
1905 my $key = $db->first_key();
1907 print "$key: " . $db->get($key) . "\n";
1908 $key = $db->next_key($key);
1911 if ($db->exists("foo")) { $db->delete("foo"); }
1915 For arrays, DBM::Deep supports all the common methods described above, and the
1916 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1917 C<unshift()> and C<splice()>.
1923 Returns the number of elements in the array. Takes no arguments.
1925 my $len = $db->length();
1929 Adds one or more elements onto the end of the array. Accepts scalars, hash
1930 refs or array refs. No return value.
1932 $db->push("foo", "bar", {});
1936 Fetches the last element in the array, and deletes it. Takes no arguments.
1937 Returns undef if array is empty. Returns the element value.
1939 my $elem = $db->pop();
1943 Fetches the first element in the array, deletes it, then shifts all the
1944 remaining elements over to take up the space. Returns the element value. This
1945 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1948 my $elem = $db->shift();
1952 Inserts one or more elements onto the beginning of the array, shifting all
1953 existing elements over to make room. Accepts scalars, hash refs or array refs.
1954 No return value. This method is not recommended with large arrays -- see
1955 <LARGE ARRAYS> below for details.
1957 $db->unshift("foo", "bar", {});
1961 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1962 -f splice> for usage -- it is too complicated to document here. This method is
1963 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1967 Here are some examples of using arrays:
1969 my $db = DBM::Deep->new(
1971 type => DBM::Deep->TYPE_ARRAY
1974 $db->push("bar", "baz");
1975 $db->unshift("foo");
1978 my $len = $db->length();
1979 print "length: $len\n"; # 4
1981 for (my $k=0; $k<$len; $k++) {
1982 print "$k: " . $db->get($k) . "\n";
1985 $db->splice(1, 2, "biz", "baf");
1987 while (my $elem = shift @$db) {
1988 print "shifted: $elem\n";
1993 Enable automatic file locking by passing a true value to the C<locking>
1994 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1996 my $db = DBM::Deep->new(
2001 This causes DBM::Deep to C<flock()> the underlying FileHandle object with exclusive
2002 mode for writes, and shared mode for reads. This is required if you have
2003 multiple processes accessing the same database file, to avoid file corruption.
2004 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
2005 NFS> below for more.
2007 =head2 EXPLICIT LOCKING
2009 You can explicitly lock a database, so it remains locked for multiple
2010 transactions. This is done by calling the C<lock()> method, and passing an
2011 optional lock mode argument (defaults to exclusive mode). This is particularly
2012 useful for things like counters, where the current value needs to be fetched,
2013 then incremented, then stored again.
2016 my $counter = $db->get("counter");
2018 $db->put("counter", $counter);
2027 You can pass C<lock()> an optional argument, which specifies which mode to use
2028 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2029 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2030 same as the constants defined in Perl's C<Fcntl> module.
2032 $db->lock( DBM::Deep->LOCK_SH );
2036 If you want to implement your own file locking scheme, be sure to create your
2037 DBM::Deep objects setting the C<volatile> option to true. This hints to DBM::Deep
2038 that the DB file may change between transactions. See L<LOW-LEVEL ACCESS>
2041 =head1 IMPORTING/EXPORTING
2043 You can import existing complex structures by calling the C<import()> method,
2044 and export an entire database into an in-memory structure using the C<export()>
2045 method. Both are examined here.
2049 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2050 walking the structure and adding keys/elements to the database as you go,
2051 simply pass a reference to the C<import()> method. This recursively adds
2052 everything to an existing DBM::Deep object for you. Here is an example:
2057 array1 => [ "elem0", "elem1", "elem2" ],
2059 subkey1 => "subvalue1",
2060 subkey2 => "subvalue2"
2064 my $db = DBM::Deep->new( "foo.db" );
2065 $db->import( $struct );
2067 print $db->{key1} . "\n"; # prints "value1"
2069 This recursively imports the entire C<$struct> object into C<$db>, including
2070 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2071 keys are merged with the existing ones, replacing if they already exist.
2072 The C<import()> method can be called on any database level (not just the base
2073 level), and works with both hash and array DB types.
2077 B<Note:> Make sure your existing structure has no circular references in it.
2078 These will cause an infinite loop when importing.
2082 Calling the C<export()> method on an existing DBM::Deep object will return
2083 a reference to a new in-memory copy of the database. The export is done
2084 recursively, so all nested hashes/arrays are all exported to standard Perl
2085 objects. Here is an example:
2087 my $db = DBM::Deep->new( "foo.db" );
2089 $db->{key1} = "value1";
2090 $db->{key2} = "value2";
2092 $db->{hash1}->{subkey1} = "subvalue1";
2093 $db->{hash1}->{subkey2} = "subvalue2";
2095 my $struct = $db->export();
2097 print $struct->{key1} . "\n"; # prints "value1"
2099 This makes a complete copy of the database in memory, and returns a reference
2100 to it. The C<export()> method can be called on any database level (not just
2101 the base level), and works with both hash and array DB types. Be careful of
2102 large databases -- you can store a lot more data in a DBM::Deep object than an
2103 in-memory Perl structure.
2107 B<Note:> Make sure your database has no circular references in it.
2108 These will cause an infinite loop when exporting.
2112 DBM::Deep has a number of hooks where you can specify your own Perl function
2113 to perform filtering on incoming or outgoing data. This is a perfect
2114 way to extend the engine, and implement things like real-time compression or
2115 encryption. Filtering applies to the base DB level, and all child hashes /
2116 arrays. Filter hooks can be specified when your DBM::Deep object is first
2117 constructed, or by calling the C<set_filter()> method at any time. There are
2118 four available filter hooks, described below:
2122 =item * filter_store_key
2124 This filter is called whenever a hash key is stored. It
2125 is passed the incoming key, and expected to return a transformed key.
2127 =item * filter_store_value
2129 This filter is called whenever a hash key or array element is stored. It
2130 is passed the incoming value, and expected to return a transformed value.
2132 =item * filter_fetch_key
2134 This filter is called whenever a hash key is fetched (i.e. via
2135 C<first_key()> or C<next_key()>). It is passed the transformed key,
2136 and expected to return the plain key.
2138 =item * filter_fetch_value
2140 This filter is called whenever a hash key or array element is fetched.
2141 It is passed the transformed value, and expected to return the plain value.
2145 Here are the two ways to setup a filter hook:
2147 my $db = DBM::Deep->new(
2149 filter_store_value => \&my_filter_store,
2150 filter_fetch_value => \&my_filter_fetch
2155 $db->set_filter( "filter_store_value", \&my_filter_store );
2156 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2158 Your filter function will be called only when dealing with SCALAR keys or
2159 values. When nested hashes and arrays are being stored/fetched, filtering
2160 is bypassed. Filters are called as static functions, passed a single SCALAR
2161 argument, and expected to return a single SCALAR value. If you want to
2162 remove a filter, set the function reference to C<undef>:
2164 $db->set_filter( "filter_store_value", undef );
2166 =head2 REAL-TIME ENCRYPTION EXAMPLE
2168 Here is a working example that uses the I<Crypt::Blowfish> module to
2169 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2170 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2171 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2174 use Crypt::Blowfish;
2177 my $cipher = Crypt::CBC->new({
2178 'key' => 'my secret key',
2179 'cipher' => 'Blowfish',
2181 'regenerate_key' => 0,
2182 'padding' => 'space',
2186 my $db = DBM::Deep->new(
2187 file => "foo-encrypt.db",
2188 filter_store_key => \&my_encrypt,
2189 filter_store_value => \&my_encrypt,
2190 filter_fetch_key => \&my_decrypt,
2191 filter_fetch_value => \&my_decrypt,
2194 $db->{key1} = "value1";
2195 $db->{key2} = "value2";
2196 print "key1: " . $db->{key1} . "\n";
2197 print "key2: " . $db->{key2} . "\n";
2203 return $cipher->encrypt( $_[0] );
2206 return $cipher->decrypt( $_[0] );
2209 =head2 REAL-TIME COMPRESSION EXAMPLE
2211 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2212 compression / decompression of keys & values with DBM::Deep Filters.
2213 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2214 more on I<Compress::Zlib>.
2219 my $db = DBM::Deep->new(
2220 file => "foo-compress.db",
2221 filter_store_key => \&my_compress,
2222 filter_store_value => \&my_compress,
2223 filter_fetch_key => \&my_decompress,
2224 filter_fetch_value => \&my_decompress,
2227 $db->{key1} = "value1";
2228 $db->{key2} = "value2";
2229 print "key1: " . $db->{key1} . "\n";
2230 print "key2: " . $db->{key2} . "\n";
2236 return Compress::Zlib::memGzip( $_[0] ) ;
2239 return Compress::Zlib::memGunzip( $_[0] ) ;
2242 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2243 actually numerical index numbers, and are not filtered.
2245 =head1 ERROR HANDLING
2247 Most DBM::Deep methods return a true value for success, and call die() on
2248 failure. You can wrap calls in an eval block to catch the die. Also, the
2249 actual error message is stored in an internal scalar, which can be fetched by
2250 calling the C<error()> method.
2252 my $db = DBM::Deep->new( "foo.db" ); # create hash
2253 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2255 print $db->error(); # prints error message
2257 You can then call C<clear_error()> to clear the current error state.
2261 If you set the C<debug> option to true when creating your DBM::Deep object,
2262 all errors are considered NON-FATAL, and dumped to STDERR. This is only
2263 for debugging purposes.
2265 =head1 LARGEFILE SUPPORT
2267 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2268 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2269 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2270 by calling the static C<set_pack()> method before you do anything else.
2272 DBM::Deep::set_pack(8, 'Q');
2274 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2275 instead of 32-bit longs. After setting these values your DB files have a
2276 theoretical maximum size of 16 XB (exabytes).
2280 B<Note:> Changing these values will B<NOT> work for existing database files.
2281 Only change this for new files, and make sure it stays set consistently
2282 throughout the file's life. If you do set these values, you can no longer
2283 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2284 back to 32-bit mode.
2288 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2289 only a 32-bit Perl. However, I have received user reports that this does
2292 =head1 LOW-LEVEL ACCESS
2294 If you require low-level access to the underlying FileHandle that DBM::Deep uses,
2295 you can call the C<fh()> method, which returns the handle:
2299 This method can be called on the root level of the datbase, or any child
2300 hashes or arrays. All levels share a I<root> structure, which contains things
2301 like the FileHandle, a reference counter, and all your options you specified
2302 when you created the object. You can get access to this root structure by
2303 calling the C<root()> method.
2305 my $root = $db->root();
2307 This is useful for changing options after the object has already been created,
2308 such as enabling/disabling locking, volatile or debug modes. You can also
2309 store your own temporary user data in this structure (be wary of name
2310 collision), which is then accessible from any child hash or array.
2312 =head1 CUSTOM DIGEST ALGORITHM
2314 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2315 keys. However you can override this, and use another algorithm (such as SHA-256)
2316 or even write your own. But please note that DBM::Deep currently expects zero
2317 collisions, so your algorithm has to be I<perfect>, so to speak.
2318 Collision detection may be introduced in a later version.
2322 You can specify a custom digest algorithm by calling the static C<set_digest()>
2323 function, passing a reference to a subroutine, and the length of the algorithm's
2324 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2325 objects. Here is a working example that uses a 256-bit hash from the
2326 I<Digest::SHA256> module. Please see
2327 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2332 my $context = Digest::SHA256::new(256);
2334 DBM::Deep::set_digest( \&my_digest, 32 );
2336 my $db = DBM::Deep->new( "foo-sha.db" );
2338 $db->{key1} = "value1";
2339 $db->{key2} = "value2";
2340 print "key1: " . $db->{key1} . "\n";
2341 print "key2: " . $db->{key2} . "\n";
2347 return substr( $context->hash($_[0]), 0, 32 );
2350 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2351 of bytes you specify in the C<set_digest()> function (in this case 32).
2353 =head1 CIRCULAR REFERENCES
2355 DBM::Deep has B<experimental> support for circular references. Meaning you
2356 can have a nested hash key or array element that points to a parent object.
2357 This relationship is stored in the DB file, and is preserved between sessions.
2360 my $db = DBM::Deep->new( "foo.db" );
2363 $db->{circle} = $db; # ref to self
2365 print $db->{foo} . "\n"; # prints "foo"
2366 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2368 One catch is, passing the object to a function that recursively walks the
2369 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2370 C<export()> methods) will result in an infinite loop. The other catch is,
2371 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2372 or C<next_key()> methods), you will get the I<target object's key>, not the
2373 ref's key. This gets even more interesting with the above example, where
2374 the I<circle> key points to the base DB object, which technically doesn't
2375 have a key. So I made DBM::Deep return "[base]" as the key name in that
2378 =head1 CAVEATS / ISSUES / BUGS
2380 This section describes all the known issues with DBM::Deep. It you have found
2381 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2383 =head2 UNUSED SPACE RECOVERY
2385 One major caveat with DBM::Deep is that space occupied by existing keys and
2386 values is not recovered when they are deleted. Meaning if you keep deleting
2387 and adding new keys, your file will continuously grow. I am working on this,
2388 but in the meantime you can call the built-in C<optimize()> method from time to
2389 time (perhaps in a crontab or something) to recover all your unused space.
2391 $db->optimize(); # returns true on success
2393 This rebuilds the ENTIRE database into a new file, then moves it on top of
2394 the original. The new file will have no unused space, thus it will take up as
2395 little disk space as possible. Please note that this operation can take
2396 a long time for large files, and you need enough disk space to temporarily hold
2397 2 copies of your DB file. The temporary file is created in the same directory
2398 as the original, named with a ".tmp" extension, and is deleted when the
2399 operation completes. Oh, and if locking is enabled, the DB is automatically
2400 locked for the entire duration of the copy.
2404 B<WARNING:> Only call optimize() on the top-level node of the database, and
2405 make sure there are no child references lying around. DBM::Deep keeps a reference
2406 counter, and if it is greater than 1, optimize() will abort and return undef.
2408 =head2 AUTOVIVIFICATION
2410 Unfortunately, autovivification doesn't work with tied hashes. This appears to
2411 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
2412 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
2413 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
2414 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
2417 $db->{foo}->{bar} = "hello";
2419 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
2420 being an empty hash. Try this instead, which works fine:
2422 $db->{foo} = { bar => "hello" };
2424 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
2425 the execution path, and Perl indeed passes an empty hash to the STORE() method.
2426 Probably a bug in Perl.
2428 =head2 FILE CORRUPTION
2430 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2431 for a 32-bit signature when opened, but other corruption in files can cause
2432 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2433 stuck in an infinite loop depending on the level of corruption. File write
2434 operations are not checked for failure (for speed), so if you happen to run
2435 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2436 be addressed in a later version of DBM::Deep.
2440 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2441 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2442 about setting up your NFS server with a locking daemon, then using lockf() to
2443 lock your files, but your milage may vary there as well. From what I
2444 understand, there is no real way to do it. However, if you need access to the
2445 underlying FileHandle in DBM::Deep for using some other kind of locking scheme like
2446 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2448 =head2 COPYING OBJECTS
2450 Beware of copying tied objects in Perl. Very strange things can happen.
2451 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2452 returns a new, blessed, tied hash or array to the same level in the DB.
2454 my $copy = $db->clone();
2458 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2459 These functions cause every element in the array to move, which can be murder
2460 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2461 a different location. This may be addressed in a later version.
2465 This section discusses DBM::Deep's speed and memory usage.
2469 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2470 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2471 multi-level hash/array support, and cross-platform FTPable files. Even so,
2472 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2473 with huge databases. Here is some test data:
2475 Adding 1,000,000 keys to new DB file...
2477 At 100 keys, avg. speed is 2,703 keys/sec
2478 At 200 keys, avg. speed is 2,642 keys/sec
2479 At 300 keys, avg. speed is 2,598 keys/sec
2480 At 400 keys, avg. speed is 2,578 keys/sec
2481 At 500 keys, avg. speed is 2,722 keys/sec
2482 At 600 keys, avg. speed is 2,628 keys/sec
2483 At 700 keys, avg. speed is 2,700 keys/sec
2484 At 800 keys, avg. speed is 2,607 keys/sec
2485 At 900 keys, avg. speed is 2,190 keys/sec
2486 At 1,000 keys, avg. speed is 2,570 keys/sec
2487 At 2,000 keys, avg. speed is 2,417 keys/sec
2488 At 3,000 keys, avg. speed is 1,982 keys/sec
2489 At 4,000 keys, avg. speed is 1,568 keys/sec
2490 At 5,000 keys, avg. speed is 1,533 keys/sec
2491 At 6,000 keys, avg. speed is 1,787 keys/sec
2492 At 7,000 keys, avg. speed is 1,977 keys/sec
2493 At 8,000 keys, avg. speed is 2,028 keys/sec
2494 At 9,000 keys, avg. speed is 2,077 keys/sec
2495 At 10,000 keys, avg. speed is 2,031 keys/sec
2496 At 20,000 keys, avg. speed is 1,970 keys/sec
2497 At 30,000 keys, avg. speed is 2,050 keys/sec
2498 At 40,000 keys, avg. speed is 2,073 keys/sec
2499 At 50,000 keys, avg. speed is 1,973 keys/sec
2500 At 60,000 keys, avg. speed is 1,914 keys/sec
2501 At 70,000 keys, avg. speed is 2,091 keys/sec
2502 At 80,000 keys, avg. speed is 2,103 keys/sec
2503 At 90,000 keys, avg. speed is 1,886 keys/sec
2504 At 100,000 keys, avg. speed is 1,970 keys/sec
2505 At 200,000 keys, avg. speed is 2,053 keys/sec
2506 At 300,000 keys, avg. speed is 1,697 keys/sec
2507 At 400,000 keys, avg. speed is 1,838 keys/sec
2508 At 500,000 keys, avg. speed is 1,941 keys/sec
2509 At 600,000 keys, avg. speed is 1,930 keys/sec
2510 At 700,000 keys, avg. speed is 1,735 keys/sec
2511 At 800,000 keys, avg. speed is 1,795 keys/sec
2512 At 900,000 keys, avg. speed is 1,221 keys/sec
2513 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2515 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2516 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2517 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2518 Run time was 12 min 3 sec.
2522 One of the great things about DBM::Deep is that it uses very little memory.
2523 Even with huge databases (1,000,000+ keys) you will not see much increased
2524 memory on your process. DBM::Deep relies solely on the filesystem for storing
2525 and fetching data. Here is output from I</usr/bin/top> before even opening a
2528 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2529 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2531 Basically the process is taking 2,716K of memory. And here is the same
2532 process after storing and fetching 1,000,000 keys:
2534 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2535 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2537 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2538 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2540 =head1 DB FILE FORMAT
2542 In case you were interested in the underlying DB file format, it is documented
2543 here in this section. You don't need to know this to use the module, it's just
2544 included for reference.
2548 DBM::Deep files always start with a 32-bit signature to identify the file type.
2549 This is at offset 0. The signature is "DPDB" in network byte order. This is
2550 checked when the file is opened.
2554 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2555 has a standard header containing the type of data, the length of data, and then
2556 the data itself. The type is a single character (1 byte), the length is a
2557 32-bit unsigned long in network byte order, and the data is, well, the data.
2558 Here is how it unfolds:
2562 Immediately after the 32-bit file signature is the I<Master Index> record.
2563 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2564 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2565 depending on how the DBM::Deep object was constructed.
2569 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2570 number). The first 8-bit char of the MD5 signature is the offset into the
2571 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2572 index element is a file offset of the next tag for the key/element in question,
2573 which is usually a I<Bucket List> tag (see below).
2577 The next tag I<could> be another index, depending on how many keys/elements
2578 exist. See L<RE-INDEXING> below for details.
2582 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2583 file offsets to where the actual data is stored. It starts with a standard
2584 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2585 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2586 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2587 When the list fills up, a I<Re-Index> operation is performed (See
2588 L<RE-INDEXING> below).
2592 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2593 index/value pair (in array mode). It starts with a standard tag header with
2594 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2595 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2596 header. The size reported in the tag header is only for the value, but then,
2597 just after the value is another size (32-bit unsigned long) and then the plain
2598 key itself. Since the value is likely to be fetched more often than the plain
2599 key, I figured it would be I<slightly> faster to store the value first.
2603 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2604 record for the nested structure, where the process begins all over again.
2608 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2609 exhausted. Then, when another key/element comes in, the list is converted to a
2610 new index record. However, this index will look at the next char in the MD5
2611 hash, and arrange new Bucket List pointers accordingly. This process is called
2612 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2613 17 (16 + new one) keys/elements are removed from the old Bucket List and
2614 inserted into the new index. Several new Bucket Lists are created in the
2615 process, as a new MD5 char from the key is being examined (it is unlikely that
2616 the keys will all share the same next char of their MD5s).
2620 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2621 when the Bucket Lists will turn into indexes, but the first round tends to
2622 happen right around 4,000 keys. You will see a I<slight> decrease in
2623 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2624 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2625 right around 900,000 keys. This process can continue nearly indefinitely --
2626 right up until the point the I<MD5> signatures start colliding with each other,
2627 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2628 getting struck by lightning while you are walking to cash in your tickets.
2629 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2630 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2631 this is 340 unodecillion, but don't quote me).
2635 When a new key/element is stored, the key (or index number) is first ran through
2636 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2637 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2638 for the first char of the signature (in this case I<b>). If it does not exist,
2639 a new I<Bucket List> is created for our key (and the next 15 future keys that
2640 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2641 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2642 this point, unless we are replacing an existing I<Bucket>), where the actual
2643 data will be stored.
2647 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2648 (or index number), then walking along the indexes. If there are enough
2649 keys/elements in this DB level, there might be nested indexes, each linked to
2650 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2651 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2652 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2653 plain key are stored.
2657 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2658 methods. In this process the indexes are walked systematically, and each key
2659 fetched in increasing MD5 order (which is why it appears random). Once the
2660 I<Bucket> is found, the value is skipped the plain key returned instead.
2661 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2662 alphabetically sorted. This only happens on an index-level -- as soon as the
2663 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2664 so it's pretty much undefined how the keys will come out -- just like Perl's
2667 =head1 CODE COVERAGE
2669 I use B<Devel::Cover> to test the code coverage of my tests, below is the B<Devel::Cover> report on this
2670 module's test suite.
2672 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2673 File stmt bran cond sub pod time total
2674 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2675 blib/lib/DBM/Deep.pm 93.9 82.4 74.7 97.9 10.5 85.7 88.0
2676 blib/lib/DBM/Deep/Array.pm 97.8 84.6 50.0 100.0 n/a 9.0 94.6
2677 blib/lib/DBM/Deep/Hash.pm 93.9 87.5 100.0 100.0 n/a 5.3 93.4
2678 Total 94.4 82.9 75.8 98.5 10.5 100.0 89.0
2679 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2683 Joseph Huckaby, L<jhuckaby@cpan.org>
2685 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2689 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2690 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2694 Copyright (c) 2002-2005 Joseph Huckaby. All Rights Reserved.
2695 This is free software, you may use it and distribute it under the
2696 same terms as Perl itself.