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; }
111 if (scalar(@_) > 1) {
113 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
117 elsif ( my $type = Scalar::Util::reftype($_[0]) ) {
118 if ( $type ne 'HASH' ) {
119 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
124 $args = { file => shift };
132 # Class constructor method for Perl OO interface.
133 # Calls tie() and returns blessed reference to tied hash or array,
134 # providing a hybrid OO/tie interface.
137 my $args = $class->_get_args( @_ );
140 # Check if we want a tied hash or array.
143 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
144 $class = 'DBM::Deep::Array';
145 require DBM::Deep::Array;
146 tie @$self, $class, %$args;
149 $class = 'DBM::Deep::Hash';
150 require DBM::Deep::Hash;
151 tie %$self, $class, %$args;
154 return bless $self, $class;
159 # Setup $self and bless into this class.
164 # These are the defaults to be optionally overridden below
167 base_offset => length(SIG_FILE),
170 foreach my $param ( keys %$self ) {
171 next unless exists $args->{$param};
172 $self->{$param} = delete $args->{$param}
175 $self->{root} = exists $args->{root}
177 : DBM::Deep::_::Root->new( $args );
179 if (!defined($self->fh)) { $self->_open(); }
186 require DBM::Deep::Hash;
187 return DBM::Deep::Hash->TIEHASH( @_ );
192 require DBM::Deep::Array;
193 return DBM::Deep::Array->TIEARRAY( @_ );
196 #XXX Unneeded now ...
202 # Open a FileHandle to the database, create if nonexistent.
203 # Make sure file signature matches DeepDB spec.
205 my $self = $_[0]->_get_self;
207 if (defined($self->fh)) { $self->_close(); }
210 # Theoretically, adding O_BINARY should remove the need for the binmode
211 # Of course, testing it is going to be ... interesting.
212 my $flags = O_RDWR | O_CREAT | O_BINARY;
214 #XXX Can the mode be anything but r+, w+, or a+??
215 #XXX ie, it has to be in read-write mode
216 #XXX So, should we verify that the mode is legitimate?
218 #XXX Maybe the mode thingy should just go away. There's no good
219 #XXX reason for it ...
220 if ( $self->root->{mode} eq 'w+' ) {
225 sysopen( $fh, $self->root->{file}, $flags )
227 $self->root->{fh} = $fh;
228 }; if ($@ ) { $self->_throw_error( "Received error: $@\n" ); }
229 if (! defined($self->fh)) {
230 return $self->_throw_error("Cannot sysopen file: " . $self->root->{file} . ": $!");
235 #XXX Can we remove this by using the right sysopen() flags?
236 # Maybe ... q.v. above
237 binmode $fh; # for win32
239 if ($self->root->{autoflush}) {
240 my $old = select $fh;
246 seek($fh, 0, SEEK_SET);
249 my $bytes_read = read( $fh, $signature, length(SIG_FILE));
252 # File is empty -- write signature and master index
255 seek($fh, 0, SEEK_SET);
257 $self->root->{end} = length(SIG_FILE);
258 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
260 my $plain_key = "[base]";
261 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
262 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
264 # Flush the filehandle
265 my $old_fh = select $fh;
275 # Check signature was valid
277 unless ($signature eq SIG_FILE) {
279 return $self->_throw_error("Signature not found -- file is not a Deep DB");
282 my @stats = stat($fh);
283 $self->root->{inode} = $stats[1];
284 $self->root->{end} = $stats[7];
287 # Get our type from master index signature
289 my $tag = $self->_load_tag($self->base_offset);
291 #XXX We probably also want to store the hash algorithm name and not assume anything
292 #XXX The cool thing would be to allow a different hashing algorithm at every level
295 return $self->_throw_error("Corrupted file, no master index record");
297 if ($self->{type} ne $tag->{signature}) {
298 return $self->_throw_error("File type mismatch");
306 # Close database FileHandle
308 my $self = $_[0]->_get_self;
309 close $self->root->{fh};
310 $self->root->{fh} = undef;
315 # Given offset, signature and content, create tag and write to disk
317 my ($self, $offset, $sig, $content) = @_;
318 my $size = length($content);
322 seek($fh, $offset, SEEK_SET);
323 print($fh $sig . pack($DATA_LENGTH_PACK, $size) . $content );
325 if ($offset == $self->root->{end}) {
326 $self->root->{end} += SIG_SIZE + $DATA_LENGTH_SIZE + $size;
332 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
339 # Given offset, load single tag and return signature, size and data
346 seek($fh, $offset, SEEK_SET);
347 if (eof $fh) { return undef; }
350 read( $fh, $sig, SIG_SIZE);
353 read( $fh, $size, $DATA_LENGTH_SIZE);
354 $size = unpack($DATA_LENGTH_PACK, $size);
357 read( $fh, $buffer, $size);
362 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
369 # Given index tag, lookup single entry in index and return .
372 my ($tag, $index) = @_;
374 my $location = unpack($LONG_PACK, substr($tag->{content}, $index * $LONG_SIZE, $LONG_SIZE) );
375 if (!$location) { return; }
377 return $self->_load_tag( $location );
382 # Adds one key/value pair to bucket list, given offset, MD5 digest of key,
383 # plain (undigested) key and value.
386 my ($tag, $md5, $plain_key, $value) = @_;
387 my $keys = $tag->{content};
391 my $is_dbm_deep = eval { $value->isa( 'DBM::Deep' ) };
392 my $internal_ref = $is_dbm_deep && ($value->root eq $self->root);
397 # Iterate through buckets, seeing if this is a new entry or a replace.
399 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
400 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
401 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
404 # Found empty bucket (end of list). Populate and exit loop.
408 $location = $internal_ref
409 ? $value->base_offset
410 : $self->root->{end};
412 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
413 print($fh $md5 . pack($LONG_PACK, $location) );
416 elsif ($md5 eq $key) {
418 # Found existing bucket with same key. Replace with new value.
423 $location = $value->base_offset;
424 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
425 print($fh $md5 . pack($LONG_PACK, $location) );
428 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
430 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
433 # If value is a hash, array, or raw value with equal or less size, we can
434 # reuse the same content area of the database. Otherwise, we have to create
435 # a new content area at the EOF.
438 my $r = Scalar::Util::reftype( $value ) || '';
439 if ( $r eq 'HASH' || $r eq 'ARRAY' ) { $actual_length = $INDEX_SIZE; }
440 else { $actual_length = length($value); }
442 if ($actual_length <= $size) {
446 $location = $self->root->{end};
447 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE, SEEK_SET);
448 print($fh pack($LONG_PACK, $location) );
456 # If this is an internal reference, return now.
457 # No need to write value or plain key
464 # If bucket didn't fit into list, split into a new index level
467 seek($fh, $tag->{ref_loc}, SEEK_SET);
468 print($fh pack($LONG_PACK, $self->root->{end}) );
470 my $index_tag = $self->_create_tag($self->root->{end}, SIG_INDEX, chr(0) x $INDEX_SIZE);
473 $keys .= $md5 . pack($LONG_PACK, 0);
475 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
476 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
478 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
479 my $num = ord(substr($key, $tag->{ch} + 1, 1));
481 if ($offsets[$num]) {
482 my $offset = $offsets[$num] + SIG_SIZE + $DATA_LENGTH_SIZE;
483 seek($fh, $offset, SEEK_SET);
485 read( $fh, $subkeys, $BUCKET_LIST_SIZE);
487 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
488 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
490 seek($fh, $offset + ($k * $BUCKET_SIZE), SEEK_SET);
491 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
497 $offsets[$num] = $self->root->{end};
498 seek($fh, $index_tag->{offset} + ($num * $LONG_SIZE), SEEK_SET);
499 print($fh pack($LONG_PACK, $self->root->{end}) );
501 my $blist_tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
503 seek($fh, $blist_tag->{offset}, SEEK_SET);
504 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
509 $location ||= $self->root->{end};
510 } # re-index bucket list
513 # Seek to content area and store signature, value and plaintext key
517 seek($fh, $location, SEEK_SET);
520 # Write signature based on content type, set content length and write actual value.
522 my $r = Scalar::Util::reftype($value) || '';
524 print($fh TYPE_HASH );
525 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
526 $content_length = $INDEX_SIZE;
528 elsif ($r eq 'ARRAY') {
529 print($fh TYPE_ARRAY );
530 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
531 $content_length = $INDEX_SIZE;
533 elsif (!defined($value)) {
534 print($fh SIG_NULL );
535 print($fh pack($DATA_LENGTH_PACK, 0) );
539 print($fh SIG_DATA );
540 print($fh pack($DATA_LENGTH_PACK, length($value)) . $value );
541 $content_length = length($value);
545 # Plain key is stored AFTER value, as keys are typically fetched less often.
547 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
550 # If value is blessed, preserve class name
552 if ( $self->root->{autobless} ) {
553 my $value_class = Scalar::Util::blessed($value);
554 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
556 # Blessed ref -- will restore later
559 print($fh pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
560 $content_length += 1;
561 $content_length += $DATA_LENGTH_SIZE + length($value_class);
565 $content_length += 1;
570 # If this is a new content area, advance EOF counter
572 if ($location == $self->root->{end}) {
573 $self->root->{end} += SIG_SIZE;
574 $self->root->{end} += $DATA_LENGTH_SIZE + $content_length;
575 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
579 # If content is a hash or array, create new child DeepDB object and
580 # pass each key or element to it.
583 my $branch = DBM::Deep->new(
585 base_offset => $location,
588 foreach my $key (keys %{$value}) {
589 #$branch->{$key} = $value->{$key};
590 $branch->STORE( $key, $value->{$key} );
593 elsif ($r eq 'ARRAY') {
594 my $branch = DBM::Deep->new(
596 base_offset => $location,
600 foreach my $element (@{$value}) {
601 #$branch->[$index] = $element;
602 $branch->STORE( $index, $element );
610 return $self->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
613 sub _get_bucket_value {
615 # Fetch single value given tag and MD5 digested key.
618 my ($tag, $md5) = @_;
619 my $keys = $tag->{content};
624 # Iterate through buckets, looking for a key match
627 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
628 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
629 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
633 # Hit end of list, no match
638 if ( $md5 ne $key ) {
643 # Found match -- seek to offset and read signature
646 seek($fh, $subloc, SEEK_SET);
647 read( $fh, $signature, SIG_SIZE);
650 # If value is a hash or array, return new DeepDB object with correct offset
652 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
653 my $obj = DBM::Deep->new(
655 base_offset => $subloc,
659 if ($self->root->{autobless}) {
661 # Skip over value and plain key to see if object needs
664 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
667 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
668 if ($size) { seek($fh, $size, SEEK_CUR); }
671 read( $fh, $bless_bit, 1);
672 if (ord($bless_bit)) {
674 # Yes, object needs to be re-blessed
677 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
678 if ($size) { read( $fh, $class_name, $size); }
679 if ($class_name) { $obj = bless( $obj, $class_name ); }
687 # Otherwise return actual value
689 elsif ($signature eq SIG_DATA) {
692 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
693 if ($size) { read( $fh, $value, $size); }
698 # Key exists, but content is null
708 # Delete single key/value pair given tag and MD5 digested key.
711 my ($tag, $md5) = @_;
712 my $keys = $tag->{content};
717 # Iterate through buckets, looking for a key match
720 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
721 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
722 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
726 # Hit end of list, no match
731 if ( $md5 ne $key ) {
736 # Matched key -- delete bucket and return
738 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
739 print($fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
740 print($fh chr(0) x $BUCKET_SIZE );
750 # Check existence of single key given tag and MD5 digested key.
753 my ($tag, $md5) = @_;
754 my $keys = $tag->{content};
757 # Iterate through buckets, looking for a key match
760 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
761 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
762 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
766 # Hit end of list, no match
771 if ( $md5 ne $key ) {
776 # Matched key -- return true
784 sub _find_bucket_list {
786 # Locate offset for bucket list, given digested key
792 # Locate offset for bucket list using digest index system
795 my $tag = $self->_load_tag($self->base_offset);
796 if (!$tag) { return; }
798 while ($tag->{signature} ne SIG_BLIST) {
799 $tag = $self->_index_lookup($tag, ord(substr($md5, $ch, 1)));
800 if (!$tag) { return; }
807 sub _traverse_index {
809 # Scan index and recursively step into deeper levels, looking for next key.
811 my ($self, $offset, $ch, $force_return_next) = @_;
812 $force_return_next = undef unless $force_return_next;
814 my $tag = $self->_load_tag( $offset );
818 if ($tag->{signature} ne SIG_BLIST) {
819 my $content = $tag->{content};
821 if ($self->{return_next}) { $start = 0; }
822 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
824 for (my $index = $start; $index < 256; $index++) {
825 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
827 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
828 if (defined($result)) { return $result; }
832 $self->{return_next} = 1;
835 elsif ($tag->{signature} eq SIG_BLIST) {
836 my $keys = $tag->{content};
837 if ($force_return_next) { $self->{return_next} = 1; }
840 # Iterate through buckets, looking for a key match
842 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
843 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
844 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
848 # End of bucket list -- return to outer loop
850 $self->{return_next} = 1;
853 elsif ($key eq $self->{prev_md5}) {
855 # Located previous key -- return next one found
857 $self->{return_next} = 1;
860 elsif ($self->{return_next}) {
862 # Seek to bucket location and skip over signature
864 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
867 # Skip over value to get to plain key
870 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
871 if ($size) { seek($fh, $size, SEEK_CUR); }
874 # Read in plain key and return as scalar
877 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
878 if ($size) { read( $fh, $plain_key, $size); }
884 $self->{return_next} = 1;
885 } # tag is a bucket list
892 # Locate next key, given digested previous one
894 my $self = $_[0]->_get_self;
896 $self->{prev_md5} = $_[1] ? $_[1] : undef;
897 $self->{return_next} = 0;
900 # If the previous key was not specifed, start at the top and
901 # return the first one found.
903 if (!$self->{prev_md5}) {
904 $self->{prev_md5} = chr(0) x $HASH_SIZE;
905 $self->{return_next} = 1;
908 return $self->_traverse_index( $self->base_offset, 0 );
913 # If db locking is set, flock() the db file. If called multiple
914 # times before unlock(), then the same number of unlocks() must
915 # be called before the lock is released.
917 my $self = $_[0]->_get_self;
919 $type = LOCK_EX unless defined $type;
921 if (!defined($self->fh)) { return; }
923 if ($self->root->{locking}) {
924 if (!$self->root->{locked}) {
925 flock($self->fh, $type);
927 # double-check file inode, in case another process
928 # has optimize()d our file while we were waiting.
929 if ((stat($self->root->{file}))[1] != $self->root->{inode}) {
930 $self->_open(); # re-open
931 flock($self->fh, $type); # re-lock
934 $self->root->{locked}++;
944 # If db locking is set, unlock the db file. See note in lock()
945 # regarding calling lock() multiple times.
947 my $self = $_[0]->_get_self;
949 if (!defined($self->fh)) { return; }
951 if ($self->root->{locking} && $self->root->{locked} > 0) {
952 $self->root->{locked}--;
953 if (!$self->root->{locked}) { flock($self->fh, LOCK_UN); }
961 #XXX These uses of ref() need verified
964 # Copy single level of keys or elements to new DB handle.
965 # Recurse for nested structures
967 my $self = $_[0]->_get_self;
970 if ($self->type eq TYPE_HASH) {
971 my $key = $self->first_key();
973 my $value = $self->get($key);
974 #XXX This doesn't work with autobless
975 if (!ref($value)) { $db_temp->{$key} = $value; }
977 my $type = $value->type;
978 if ($type eq TYPE_HASH) { $db_temp->{$key} = {}; }
979 else { $db_temp->{$key} = []; }
980 $value->_copy_node( $db_temp->{$key} );
982 $key = $self->next_key($key);
986 my $length = $self->length();
987 for (my $index = 0; $index < $length; $index++) {
988 my $value = $self->get($index);
989 if (!ref($value)) { $db_temp->[$index] = $value; }
990 #XXX NO tests for this code
992 my $type = $value->type;
993 if ($type eq TYPE_HASH) { $db_temp->[$index] = {}; }
994 else { $db_temp->[$index] = []; }
995 $value->_copy_node( $db_temp->[$index] );
1003 # Recursively export into standard Perl hashes and arrays.
1005 my $self = $_[0]->_get_self;
1008 if ($self->type eq TYPE_HASH) { $temp = {}; }
1009 elsif ($self->type eq TYPE_ARRAY) { $temp = []; }
1012 $self->_copy_node( $temp );
1020 # Recursively import Perl hash/array structure
1022 #XXX This use of ref() seems to be ok
1023 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
1025 my $self = $_[0]->_get_self;
1028 #XXX This use of ref() seems to be ok
1029 if (!ref($struct)) {
1031 # struct is not a reference, so just import based on our type
1035 if ($self->type eq TYPE_HASH) { $struct = {@_}; }
1036 elsif ($self->type eq TYPE_ARRAY) { $struct = [@_]; }
1039 my $r = Scalar::Util::reftype($struct) || '';
1040 if ($r eq "HASH" && $self->type eq TYPE_HASH) {
1041 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
1043 elsif ($r eq "ARRAY" && $self->type eq TYPE_ARRAY) {
1044 $self->push( @$struct );
1047 return $self->_throw_error("Cannot import: type mismatch");
1055 # Rebuild entire database into new file, then move
1056 # it back on top of original.
1058 my $self = $_[0]->_get_self;
1060 #XXX Need to create a new test for this
1061 # if ($self->root->{links} > 1) {
1062 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
1065 my $db_temp = DBM::Deep->new(
1066 file => $self->root->{file} . '.tmp',
1070 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
1074 $self->_copy_node( $db_temp );
1078 # Attempt to copy user, group and permissions over to new file
1080 my @stats = stat($self->fh);
1081 my $perms = $stats[2] & 07777;
1082 my $uid = $stats[4];
1083 my $gid = $stats[5];
1084 chown( $uid, $gid, $self->root->{file} . '.tmp' );
1085 chmod( $perms, $self->root->{file} . '.tmp' );
1087 # q.v. perlport for more information on this variable
1088 if ( $^O eq 'MSWin32' ) {
1090 # Potential race condition when optmizing on Win32 with locking.
1091 # The Windows filesystem requires that the filehandle be closed
1092 # before it is overwritten with rename(). This could be redone
1099 if (!rename $self->root->{file} . '.tmp', $self->root->{file}) {
1100 unlink $self->root->{file} . '.tmp';
1102 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
1114 # Make copy of object and return
1116 my $self = $_[0]->_get_self;
1118 return DBM::Deep->new(
1119 type => $self->type,
1120 base_offset => $self->base_offset,
1126 my %is_legal_filter = map {
1129 store_key store_value
1130 fetch_key fetch_value
1135 # Setup filter function for storing or fetching the key or value
1137 my $self = $_[0]->_get_self;
1138 my $type = lc $_[1];
1139 my $func = $_[2] ? $_[2] : undef;
1141 if ( $is_legal_filter{$type} ) {
1142 $self->root->{"filter_$type"} = $func;
1156 # Get access to the root structure
1158 my $self = $_[0]->_get_self;
1159 return $self->{root};
1164 # Get access to the raw FileHandle
1166 #XXX It will be useful, though, when we split out HASH and ARRAY
1167 my $self = $_[0]->_get_self;
1168 return $self->root->{fh};
1173 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
1175 my $self = $_[0]->_get_self;
1176 return $self->{type};
1181 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
1183 my $self = $_[0]->_get_self;
1184 return $self->{base_offset};
1189 # Get last error string, or undef if no error
1192 #? ( _get_self($_[0])->{root}->{error} or undef )
1193 ? ( $_[0]->_get_self->{root}->{error} or undef )
1203 # Store error string in self
1205 my $self = $_[0]->_get_self;
1206 my $error_text = $_[1];
1208 if ( Scalar::Util::blessed $self ) {
1209 $self->root->{error} = $error_text;
1211 unless ($self->root->{debug}) {
1212 die "DBM::Deep: $error_text\n";
1215 warn "DBM::Deep: $error_text\n";
1219 die "DBM::Deep: $error_text\n";
1227 my $self = $_[0]->_get_self;
1229 undef $self->root->{error};
1234 # Precalculate index, bucket and bucket list sizes
1237 #XXX I don't like this ...
1238 set_pack() unless defined $LONG_SIZE;
1240 $INDEX_SIZE = 256 * $LONG_SIZE;
1241 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
1242 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
1247 # Set pack/unpack modes (see file header for more)
1249 my ($long_s, $long_p, $data_s, $data_p) = @_;
1251 $LONG_SIZE = $long_s ? $long_s : 4;
1252 $LONG_PACK = $long_p ? $long_p : 'N';
1254 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
1255 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
1262 # Set key digest function (default is MD5)
1264 my ($digest_func, $hash_size) = @_;
1266 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
1267 $HASH_SIZE = $hash_size ? $hash_size : 16;
1273 # tie() methods (hashes and arrays)
1278 # Store single hash key/value or array element in database.
1280 my $self = $_[0]->_get_self;
1283 # User may be storing a hash, in which case we do not want it run
1284 # through the filtering system
1285 my $value = ($self->root->{filter_store_value} && !ref($_[2]))
1286 ? $self->root->{filter_store_value}->($_[2])
1289 my $md5 = $DIGEST_FUNC->($key);
1292 # Make sure file is open
1294 if (!defined($self->fh) && !$self->_open()) {
1300 # Request exclusive lock for writing
1302 $self->lock( LOCK_EX );
1307 # If locking is enabled, set 'end' parameter again, in case another
1308 # DB instance appended to our file while we were unlocked.
1310 if ($self->root->{locking} || $self->root->{volatile}) {
1311 $self->root->{end} = (stat($fh))[7];
1315 # Locate offset for bucket list using digest index system
1317 my $tag = $self->_load_tag($self->base_offset);
1319 $tag = $self->_create_tag($self->base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1323 while ($tag->{signature} ne SIG_BLIST) {
1324 my $num = ord(substr($md5, $ch, 1));
1325 my $new_tag = $self->_index_lookup($tag, $num);
1327 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1328 seek($fh, $ref_loc, SEEK_SET);
1329 print($fh pack($LONG_PACK, $self->root->{end}) );
1331 $tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1332 $tag->{ref_loc} = $ref_loc;
1337 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1339 $tag->{ref_loc} = $ref_loc;
1346 # Add key/value to bucket list
1348 my $result = $self->_add_bucket( $tag, $md5, $key, $value );
1357 # Fetch single value or element given plain key or array index
1359 my $self = shift->_get_self;
1363 # Make sure file is open
1365 if (!defined($self->fh)) { $self->_open(); }
1367 my $md5 = $DIGEST_FUNC->($key);
1370 # Request shared lock for reading
1372 $self->lock( LOCK_SH );
1374 my $tag = $self->_find_bucket_list( $md5 );
1381 # Get value from bucket list
1383 my $result = $self->_get_bucket_value( $tag, $md5 );
1387 #XXX What is ref() checking here?
1388 return ($result && !ref($result) && $self->root->{filter_fetch_value})
1389 ? $self->root->{filter_fetch_value}->($result)
1395 # Delete single key/value pair or element given plain key or array index
1397 my $self = $_[0]->_get_self;
1400 my $md5 = $DIGEST_FUNC->($key);
1403 # Make sure file is open
1405 if (!defined($self->fh)) { $self->_open(); }
1408 # Request exclusive lock for writing
1410 $self->lock( LOCK_EX );
1412 my $tag = $self->_find_bucket_list( $md5 );
1421 my $value = $self->_get_bucket_value( $tag, $md5 );
1422 if ($value && !ref($value) && $self->root->{filter_fetch_value}) {
1423 $value = $self->root->{filter_fetch_value}->($value);
1426 my $result = $self->_delete_bucket( $tag, $md5 );
1429 # If this object is an array and the key deleted was on the end of the stack,
1430 # decrement the length variable.
1440 # Check if a single key or element exists given plain key or array index
1442 my $self = $_[0]->_get_self;
1445 my $md5 = $DIGEST_FUNC->($key);
1448 # Make sure file is open
1450 if (!defined($self->fh)) { $self->_open(); }
1453 # Request shared lock for reading
1455 $self->lock( LOCK_SH );
1457 my $tag = $self->_find_bucket_list( $md5 );
1460 # For some reason, the built-in exists() function returns '' for false
1468 # Check if bucket exists and return 1 or ''
1470 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1479 # Clear all keys from hash, or all elements from array.
1481 my $self = $_[0]->_get_self;
1484 # Make sure file is open
1486 if (!defined($self->fh)) { $self->_open(); }
1489 # Request exclusive lock for writing
1491 $self->lock( LOCK_EX );
1495 seek($fh, $self->base_offset, SEEK_SET);
1501 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
1509 # Public method aliases
1511 sub put { (shift)->STORE( @_ ) }
1512 sub store { (shift)->STORE( @_ ) }
1513 sub get { (shift)->FETCH( @_ ) }
1514 sub fetch { (shift)->FETCH( @_ ) }
1515 sub delete { (shift)->DELETE( @_ ) }
1516 sub exists { (shift)->EXISTS( @_ ) }
1517 sub clear { (shift)->CLEAR( @_ ) }
1519 package DBM::Deep::_::Root;
1534 filter_store_key => undef,
1535 filter_store_value => undef,
1536 filter_fetch_key => undef,
1537 filter_fetch_value => undef,
1548 return unless $self;
1550 close $self->{fh} if $self->{fh};
1561 DBM::Deep - A pure perl multi-level hash/array DBM
1566 my $db = DBM::Deep->new( "foo.db" );
1568 $db->{key} = 'value'; # tie() style
1571 $db->put('key', 'value'); # OO style
1572 print $db->get('key');
1574 # true multi-level support
1575 $db->{my_complex} = [
1576 'hello', { perl => 'rules' },
1581 A unique flat-file database module, written in pure perl. True
1582 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1583 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1584 handle millions of keys and unlimited hash levels without significant
1585 slow-down. Written from the ground-up in pure perl -- this is NOT a
1586 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1587 Mac OS X and Windows.
1591 Hopefully you are using CPAN's excellent Perl module, which will download
1592 and install the module for you. If not, get the tarball, and run these
1604 Construction can be done OO-style (which is the recommended way), or using
1605 Perl's tie() function. Both are examined here.
1607 =head2 OO CONSTRUCTION
1609 The recommended way to construct a DBM::Deep object is to use the new()
1610 method, which gets you a blessed, tied hash or array reference.
1612 my $db = DBM::Deep->new( "foo.db" );
1614 This opens a new database handle, mapped to the file "foo.db". If this
1615 file does not exist, it will automatically be created. DB files are
1616 opened in "r+" (read/write) mode, and the type of object returned is a
1617 hash, unless otherwise specified (see L<OPTIONS> below).
1621 You can pass a number of options to the constructor to specify things like
1622 locking, autoflush, etc. This is done by passing an inline hash:
1624 my $db = DBM::Deep->new(
1630 Notice that the filename is now specified I<inside> the hash with
1631 the "file" parameter, as opposed to being the sole argument to the
1632 constructor. This is required if any options are specified.
1633 See L<OPTIONS> below for the complete list.
1637 You can also start with an array instead of a hash. For this, you must
1638 specify the C<type> parameter:
1640 my $db = DBM::Deep->new(
1642 type => DBM::Deep->TYPE_ARRAY
1645 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1646 a new DB file. If you create a DBM::Deep object with an existing file, the
1647 C<type> will be loaded from the file header, and ignored if it is passed
1650 =head2 TIE CONSTRUCTION
1652 Alternatively, you can create a DBM::Deep handle by using Perl's built-in
1653 tie() function. This is not ideal, because you get only a basic, tied hash
1654 (or array) which is not blessed, so you can't call any functions on it.
1657 tie %hash, "DBM::Deep", "foo.db";
1660 tie @array, "DBM::Deep", "bar.db";
1662 As with the OO constructor, you can replace the DB filename parameter with
1663 a hash containing one or more options (see L<OPTIONS> just below for the
1666 tie %hash, "DBM::Deep", {
1674 There are a number of options that can be passed in when constructing your
1675 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1681 Filename of the DB file to link the handle to. You can pass a full absolute
1682 filesystem path, partial path, or a plain filename if the file is in the
1683 current working directory. This is a required parameter.
1687 File open mode (read-only, read-write, etc.) string passed to Perl's FileHandle
1688 module. This is an optional parameter, and defaults to "r+" (read/write).
1689 B<Note:> If the default (r+) mode is selected, the file will also be auto-
1690 created if it doesn't exist.
1694 This parameter specifies what type of object to create, a hash or array. Use
1695 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1696 This only takes effect when beginning a new file. This is an optional
1697 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1701 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1702 function to lock the database in exclusive mode for writes, and shared mode for
1703 reads. Pass any true value to enable. This affects the base DB handle I<and
1704 any child hashes or arrays> that use the same DB file. This is an optional
1705 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1709 Specifies whether autoflush is to be enabled on the underlying FileHandle.
1710 This obviously slows down write operations, but is required if you may have
1711 multiple processes accessing the same DB file (also consider enable I<locking>
1712 or at least I<volatile>). Pass any true value to enable. This is an optional
1713 parameter, and defaults to 0 (disabled).
1717 If I<volatile> mode is enabled, DBM::Deep will stat() the DB file before each
1718 STORE() operation. This is required if an outside force may change the size of
1719 the file between transactions. Locking also implicitly enables volatile. This
1720 is useful if you want to use a different locking system or write your own. Pass
1721 any true value to enable. This is an optional parameter, and defaults to 0
1726 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1727 restore them when fetched. This is an B<experimental> feature, and does have
1728 side-effects. Basically, when hashes are re-blessed into their original
1729 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1730 able to call any DBM::Deep methods on them. You have been warned.
1731 This is an optional parameter, and defaults to 0 (disabled).
1735 See L<FILTERS> below.
1739 Setting I<debug> mode will make all errors non-fatal, dump them out to
1740 STDERR, and continue on. This is for debugging purposes only, and probably
1741 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1745 Instead of passing a file path, you can instead pass a handle to an pre-opened
1746 filehandle. Note: Beware of using the magick *DATA handle, as this actually
1747 contains your entire Perl script, as well as the data following the __DATA__
1748 marker. This will not work, because DBM::Deep uses absolute seek()s into the
1749 file. Instead, consider reading *DATA into an IO::Scalar handle, then passing
1754 =head1 TIE INTERFACE
1756 With DBM::Deep you can access your databases using Perl's standard hash/array
1757 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can treat
1758 them as such. DBM::Deep will intercept all reads/writes and direct them to the right
1759 place -- the DB file. This has nothing to do with the L<TIE CONSTRUCTION>
1760 section above. This simply tells you how to use DBM::Deep using regular hashes
1761 and arrays, rather than calling functions like C<get()> and C<put()> (although those
1762 work too). It is entirely up to you how to want to access your databases.
1766 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1767 or even nested hashes (or arrays) using standard Perl syntax:
1769 my $db = DBM::Deep->new( "foo.db" );
1771 $db->{mykey} = "myvalue";
1773 $db->{myhash}->{subkey} = "subvalue";
1775 print $db->{myhash}->{subkey} . "\n";
1777 You can even step through hash keys using the normal Perl C<keys()> function:
1779 foreach my $key (keys %$db) {
1780 print "$key: " . $db->{$key} . "\n";
1783 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1784 pushes them onto an array, all before the loop even begins. If you have an
1785 extra large hash, this may exhaust Perl's memory. Instead, consider using
1786 Perl's C<each()> function, which pulls keys/values one at a time, using very
1789 while (my ($key, $value) = each %$db) {
1790 print "$key: $value\n";
1793 Please note that when using C<each()>, you should always pass a direct
1794 hash reference, not a lookup. Meaning, you should B<never> do this:
1797 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1799 This causes an infinite loop, because for each iteration, Perl is calling
1800 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1801 it effectively keeps returning the first key over and over again. Instead,
1802 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1806 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1807 reference. This includes inserting, removing and manipulating elements,
1808 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1809 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1810 or simply be a nested array reference inside a hash. Example:
1812 my $db = DBM::Deep->new(
1813 file => "foo-array.db",
1814 type => DBM::Deep->TYPE_ARRAY
1818 push @$db, "bar", "baz";
1819 unshift @$db, "bah";
1821 my $last_elem = pop @$db; # baz
1822 my $first_elem = shift @$db; # bah
1823 my $second_elem = $db->[1]; # bar
1825 my $num_elements = scalar @$db;
1829 In addition to the I<tie()> interface, you can also use a standard OO interface
1830 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1831 array) has its own methods, but both types share the following common methods:
1832 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1838 Stores a new hash key/value pair, or sets an array element value. Takes two
1839 arguments, the hash key or array index, and the new value. The value can be
1840 a scalar, hash ref or array ref. Returns true on success, false on failure.
1842 $db->put("foo", "bar"); # for hashes
1843 $db->put(1, "bar"); # for arrays
1847 Fetches the value of a hash key or array element. Takes one argument: the hash
1848 key or array index. Returns a scalar, hash ref or array ref, depending on the
1851 my $value = $db->get("foo"); # for hashes
1852 my $value = $db->get(1); # for arrays
1856 Checks if a hash key or array index exists. Takes one argument: the hash key
1857 or array index. Returns true if it exists, false if not.
1859 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1860 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1864 Deletes one hash key/value pair or array element. Takes one argument: the hash
1865 key or array index. Returns true on success, false if not found. For arrays,
1866 the remaining elements located after the deleted element are NOT moved over.
1867 The deleted element is essentially just undefined, which is exactly how Perl's
1868 internal arrays work. Please note that the space occupied by the deleted
1869 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1870 below for details and workarounds.
1872 $db->delete("foo"); # for hashes
1873 $db->delete(1); # for arrays
1877 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1878 value. Please note that the space occupied by the deleted keys/values or
1879 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1880 details and workarounds.
1882 $db->clear(); # hashes or arrays
1888 For hashes, DBM::Deep supports all the common methods described above, and the
1889 following additional methods: C<first_key()> and C<next_key()>.
1895 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1896 fetched in an undefined order (which appears random). Takes no arguments,
1897 returns the key as a scalar value.
1899 my $key = $db->first_key();
1903 Returns the "next" key in the hash, given the previous one as the sole argument.
1904 Returns undef if there are no more keys to be fetched.
1906 $key = $db->next_key($key);
1910 Here are some examples of using hashes:
1912 my $db = DBM::Deep->new( "foo.db" );
1914 $db->put("foo", "bar");
1915 print "foo: " . $db->get("foo") . "\n";
1917 $db->put("baz", {}); # new child hash ref
1918 $db->get("baz")->put("buz", "biz");
1919 print "buz: " . $db->get("baz")->get("buz") . "\n";
1921 my $key = $db->first_key();
1923 print "$key: " . $db->get($key) . "\n";
1924 $key = $db->next_key($key);
1927 if ($db->exists("foo")) { $db->delete("foo"); }
1931 For arrays, DBM::Deep supports all the common methods described above, and the
1932 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1933 C<unshift()> and C<splice()>.
1939 Returns the number of elements in the array. Takes no arguments.
1941 my $len = $db->length();
1945 Adds one or more elements onto the end of the array. Accepts scalars, hash
1946 refs or array refs. No return value.
1948 $db->push("foo", "bar", {});
1952 Fetches the last element in the array, and deletes it. Takes no arguments.
1953 Returns undef if array is empty. Returns the element value.
1955 my $elem = $db->pop();
1959 Fetches the first element in the array, deletes it, then shifts all the
1960 remaining elements over to take up the space. Returns the element value. This
1961 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1964 my $elem = $db->shift();
1968 Inserts one or more elements onto the beginning of the array, shifting all
1969 existing elements over to make room. Accepts scalars, hash refs or array refs.
1970 No return value. This method is not recommended with large arrays -- see
1971 <LARGE ARRAYS> below for details.
1973 $db->unshift("foo", "bar", {});
1977 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1978 -f splice> for usage -- it is too complicated to document here. This method is
1979 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1983 Here are some examples of using arrays:
1985 my $db = DBM::Deep->new(
1987 type => DBM::Deep->TYPE_ARRAY
1990 $db->push("bar", "baz");
1991 $db->unshift("foo");
1994 my $len = $db->length();
1995 print "length: $len\n"; # 4
1997 for (my $k=0; $k<$len; $k++) {
1998 print "$k: " . $db->get($k) . "\n";
2001 $db->splice(1, 2, "biz", "baf");
2003 while (my $elem = shift @$db) {
2004 print "shifted: $elem\n";
2009 Enable automatic file locking by passing a true value to the C<locking>
2010 parameter when constructing your DBM::Deep object (see L<SETUP> above).
2012 my $db = DBM::Deep->new(
2017 This causes DBM::Deep to C<flock()> the underlying FileHandle object with exclusive
2018 mode for writes, and shared mode for reads. This is required if you have
2019 multiple processes accessing the same database file, to avoid file corruption.
2020 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
2021 NFS> below for more.
2023 =head2 EXPLICIT LOCKING
2025 You can explicitly lock a database, so it remains locked for multiple
2026 transactions. This is done by calling the C<lock()> method, and passing an
2027 optional lock mode argument (defaults to exclusive mode). This is particularly
2028 useful for things like counters, where the current value needs to be fetched,
2029 then incremented, then stored again.
2032 my $counter = $db->get("counter");
2034 $db->put("counter", $counter);
2043 You can pass C<lock()> an optional argument, which specifies which mode to use
2044 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2045 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2046 same as the constants defined in Perl's C<Fcntl> module.
2048 $db->lock( DBM::Deep->LOCK_SH );
2052 If you want to implement your own file locking scheme, be sure to create your
2053 DBM::Deep objects setting the C<volatile> option to true. This hints to DBM::Deep
2054 that the DB file may change between transactions. See L<LOW-LEVEL ACCESS>
2057 =head1 IMPORTING/EXPORTING
2059 You can import existing complex structures by calling the C<import()> method,
2060 and export an entire database into an in-memory structure using the C<export()>
2061 method. Both are examined here.
2065 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2066 walking the structure and adding keys/elements to the database as you go,
2067 simply pass a reference to the C<import()> method. This recursively adds
2068 everything to an existing DBM::Deep object for you. Here is an example:
2073 array1 => [ "elem0", "elem1", "elem2" ],
2075 subkey1 => "subvalue1",
2076 subkey2 => "subvalue2"
2080 my $db = DBM::Deep->new( "foo.db" );
2081 $db->import( $struct );
2083 print $db->{key1} . "\n"; # prints "value1"
2085 This recursively imports the entire C<$struct> object into C<$db>, including
2086 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2087 keys are merged with the existing ones, replacing if they already exist.
2088 The C<import()> method can be called on any database level (not just the base
2089 level), and works with both hash and array DB types.
2093 B<Note:> Make sure your existing structure has no circular references in it.
2094 These will cause an infinite loop when importing.
2098 Calling the C<export()> method on an existing DBM::Deep object will return
2099 a reference to a new in-memory copy of the database. The export is done
2100 recursively, so all nested hashes/arrays are all exported to standard Perl
2101 objects. Here is an example:
2103 my $db = DBM::Deep->new( "foo.db" );
2105 $db->{key1} = "value1";
2106 $db->{key2} = "value2";
2108 $db->{hash1}->{subkey1} = "subvalue1";
2109 $db->{hash1}->{subkey2} = "subvalue2";
2111 my $struct = $db->export();
2113 print $struct->{key1} . "\n"; # prints "value1"
2115 This makes a complete copy of the database in memory, and returns a reference
2116 to it. The C<export()> method can be called on any database level (not just
2117 the base level), and works with both hash and array DB types. Be careful of
2118 large databases -- you can store a lot more data in a DBM::Deep object than an
2119 in-memory Perl structure.
2123 B<Note:> Make sure your database has no circular references in it.
2124 These will cause an infinite loop when exporting.
2128 DBM::Deep has a number of hooks where you can specify your own Perl function
2129 to perform filtering on incoming or outgoing data. This is a perfect
2130 way to extend the engine, and implement things like real-time compression or
2131 encryption. Filtering applies to the base DB level, and all child hashes /
2132 arrays. Filter hooks can be specified when your DBM::Deep object is first
2133 constructed, or by calling the C<set_filter()> method at any time. There are
2134 four available filter hooks, described below:
2138 =item * filter_store_key
2140 This filter is called whenever a hash key is stored. It
2141 is passed the incoming key, and expected to return a transformed key.
2143 =item * filter_store_value
2145 This filter is called whenever a hash key or array element is stored. It
2146 is passed the incoming value, and expected to return a transformed value.
2148 =item * filter_fetch_key
2150 This filter is called whenever a hash key is fetched (i.e. via
2151 C<first_key()> or C<next_key()>). It is passed the transformed key,
2152 and expected to return the plain key.
2154 =item * filter_fetch_value
2156 This filter is called whenever a hash key or array element is fetched.
2157 It is passed the transformed value, and expected to return the plain value.
2161 Here are the two ways to setup a filter hook:
2163 my $db = DBM::Deep->new(
2165 filter_store_value => \&my_filter_store,
2166 filter_fetch_value => \&my_filter_fetch
2171 $db->set_filter( "filter_store_value", \&my_filter_store );
2172 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2174 Your filter function will be called only when dealing with SCALAR keys or
2175 values. When nested hashes and arrays are being stored/fetched, filtering
2176 is bypassed. Filters are called as static functions, passed a single SCALAR
2177 argument, and expected to return a single SCALAR value. If you want to
2178 remove a filter, set the function reference to C<undef>:
2180 $db->set_filter( "filter_store_value", undef );
2182 =head2 REAL-TIME ENCRYPTION EXAMPLE
2184 Here is a working example that uses the I<Crypt::Blowfish> module to
2185 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2186 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2187 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2190 use Crypt::Blowfish;
2193 my $cipher = Crypt::CBC->new({
2194 'key' => 'my secret key',
2195 'cipher' => 'Blowfish',
2197 'regenerate_key' => 0,
2198 'padding' => 'space',
2202 my $db = DBM::Deep->new(
2203 file => "foo-encrypt.db",
2204 filter_store_key => \&my_encrypt,
2205 filter_store_value => \&my_encrypt,
2206 filter_fetch_key => \&my_decrypt,
2207 filter_fetch_value => \&my_decrypt,
2210 $db->{key1} = "value1";
2211 $db->{key2} = "value2";
2212 print "key1: " . $db->{key1} . "\n";
2213 print "key2: " . $db->{key2} . "\n";
2219 return $cipher->encrypt( $_[0] );
2222 return $cipher->decrypt( $_[0] );
2225 =head2 REAL-TIME COMPRESSION EXAMPLE
2227 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2228 compression / decompression of keys & values with DBM::Deep Filters.
2229 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2230 more on I<Compress::Zlib>.
2235 my $db = DBM::Deep->new(
2236 file => "foo-compress.db",
2237 filter_store_key => \&my_compress,
2238 filter_store_value => \&my_compress,
2239 filter_fetch_key => \&my_decompress,
2240 filter_fetch_value => \&my_decompress,
2243 $db->{key1} = "value1";
2244 $db->{key2} = "value2";
2245 print "key1: " . $db->{key1} . "\n";
2246 print "key2: " . $db->{key2} . "\n";
2252 return Compress::Zlib::memGzip( $_[0] ) ;
2255 return Compress::Zlib::memGunzip( $_[0] ) ;
2258 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2259 actually numerical index numbers, and are not filtered.
2261 =head1 ERROR HANDLING
2263 Most DBM::Deep methods return a true value for success, and call die() on
2264 failure. You can wrap calls in an eval block to catch the die. Also, the
2265 actual error message is stored in an internal scalar, which can be fetched by
2266 calling the C<error()> method.
2268 my $db = DBM::Deep->new( "foo.db" ); # create hash
2269 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2271 print $db->error(); # prints error message
2273 You can then call C<clear_error()> to clear the current error state.
2277 If you set the C<debug> option to true when creating your DBM::Deep object,
2278 all errors are considered NON-FATAL, and dumped to STDERR. This is only
2279 for debugging purposes.
2281 =head1 LARGEFILE SUPPORT
2283 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2284 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2285 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2286 by calling the static C<set_pack()> method before you do anything else.
2288 DBM::Deep::set_pack(8, 'Q');
2290 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2291 instead of 32-bit longs. After setting these values your DB files have a
2292 theoretical maximum size of 16 XB (exabytes).
2296 B<Note:> Changing these values will B<NOT> work for existing database files.
2297 Only change this for new files, and make sure it stays set consistently
2298 throughout the file's life. If you do set these values, you can no longer
2299 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2300 back to 32-bit mode.
2304 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2305 only a 32-bit Perl. However, I have received user reports that this does
2308 =head1 LOW-LEVEL ACCESS
2310 If you require low-level access to the underlying FileHandle that DBM::Deep uses,
2311 you can call the C<fh()> method, which returns the handle:
2315 This method can be called on the root level of the datbase, or any child
2316 hashes or arrays. All levels share a I<root> structure, which contains things
2317 like the FileHandle, a reference counter, and all your options you specified
2318 when you created the object. You can get access to this root structure by
2319 calling the C<root()> method.
2321 my $root = $db->root();
2323 This is useful for changing options after the object has already been created,
2324 such as enabling/disabling locking, volatile or debug modes. You can also
2325 store your own temporary user data in this structure (be wary of name
2326 collision), which is then accessible from any child hash or array.
2328 =head1 CUSTOM DIGEST ALGORITHM
2330 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2331 keys. However you can override this, and use another algorithm (such as SHA-256)
2332 or even write your own. But please note that DBM::Deep currently expects zero
2333 collisions, so your algorithm has to be I<perfect>, so to speak.
2334 Collision detection may be introduced in a later version.
2338 You can specify a custom digest algorithm by calling the static C<set_digest()>
2339 function, passing a reference to a subroutine, and the length of the algorithm's
2340 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2341 objects. Here is a working example that uses a 256-bit hash from the
2342 I<Digest::SHA256> module. Please see
2343 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2348 my $context = Digest::SHA256::new(256);
2350 DBM::Deep::set_digest( \&my_digest, 32 );
2352 my $db = DBM::Deep->new( "foo-sha.db" );
2354 $db->{key1} = "value1";
2355 $db->{key2} = "value2";
2356 print "key1: " . $db->{key1} . "\n";
2357 print "key2: " . $db->{key2} . "\n";
2363 return substr( $context->hash($_[0]), 0, 32 );
2366 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2367 of bytes you specify in the C<set_digest()> function (in this case 32).
2369 =head1 CIRCULAR REFERENCES
2371 DBM::Deep has B<experimental> support for circular references. Meaning you
2372 can have a nested hash key or array element that points to a parent object.
2373 This relationship is stored in the DB file, and is preserved between sessions.
2376 my $db = DBM::Deep->new( "foo.db" );
2379 $db->{circle} = $db; # ref to self
2381 print $db->{foo} . "\n"; # prints "foo"
2382 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2384 One catch is, passing the object to a function that recursively walks the
2385 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2386 C<export()> methods) will result in an infinite loop. The other catch is,
2387 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2388 or C<next_key()> methods), you will get the I<target object's key>, not the
2389 ref's key. This gets even more interesting with the above example, where
2390 the I<circle> key points to the base DB object, which technically doesn't
2391 have a key. So I made DBM::Deep return "[base]" as the key name in that
2394 =head1 CAVEATS / ISSUES / BUGS
2396 This section describes all the known issues with DBM::Deep. It you have found
2397 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2399 =head2 UNUSED SPACE RECOVERY
2401 One major caveat with DBM::Deep is that space occupied by existing keys and
2402 values is not recovered when they are deleted. Meaning if you keep deleting
2403 and adding new keys, your file will continuously grow. I am working on this,
2404 but in the meantime you can call the built-in C<optimize()> method from time to
2405 time (perhaps in a crontab or something) to recover all your unused space.
2407 $db->optimize(); # returns true on success
2409 This rebuilds the ENTIRE database into a new file, then moves it on top of
2410 the original. The new file will have no unused space, thus it will take up as
2411 little disk space as possible. Please note that this operation can take
2412 a long time for large files, and you need enough disk space to temporarily hold
2413 2 copies of your DB file. The temporary file is created in the same directory
2414 as the original, named with a ".tmp" extension, and is deleted when the
2415 operation completes. Oh, and if locking is enabled, the DB is automatically
2416 locked for the entire duration of the copy.
2420 B<WARNING:> Only call optimize() on the top-level node of the database, and
2421 make sure there are no child references lying around. DBM::Deep keeps a reference
2422 counter, and if it is greater than 1, optimize() will abort and return undef.
2424 =head2 AUTOVIVIFICATION
2426 Unfortunately, autovivification doesn't work with tied hashes. This appears to
2427 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
2428 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
2429 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
2430 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
2433 $db->{foo}->{bar} = "hello";
2435 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
2436 being an empty hash. Try this instead, which works fine:
2438 $db->{foo} = { bar => "hello" };
2440 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
2441 the execution path, and Perl indeed passes an empty hash to the STORE() method.
2442 Probably a bug in Perl.
2444 =head2 FILE CORRUPTION
2446 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2447 for a 32-bit signature when opened, but other corruption in files can cause
2448 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2449 stuck in an infinite loop depending on the level of corruption. File write
2450 operations are not checked for failure (for speed), so if you happen to run
2451 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2452 be addressed in a later version of DBM::Deep.
2456 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2457 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2458 about setting up your NFS server with a locking daemon, then using lockf() to
2459 lock your files, but your milage may vary there as well. From what I
2460 understand, there is no real way to do it. However, if you need access to the
2461 underlying FileHandle in DBM::Deep for using some other kind of locking scheme like
2462 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2464 =head2 COPYING OBJECTS
2466 Beware of copying tied objects in Perl. Very strange things can happen.
2467 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2468 returns a new, blessed, tied hash or array to the same level in the DB.
2470 my $copy = $db->clone();
2474 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2475 These functions cause every element in the array to move, which can be murder
2476 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2477 a different location. This may be addressed in a later version.
2481 This section discusses DBM::Deep's speed and memory usage.
2485 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2486 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2487 multi-level hash/array support, and cross-platform FTPable files. Even so,
2488 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2489 with huge databases. Here is some test data:
2491 Adding 1,000,000 keys to new DB file...
2493 At 100 keys, avg. speed is 2,703 keys/sec
2494 At 200 keys, avg. speed is 2,642 keys/sec
2495 At 300 keys, avg. speed is 2,598 keys/sec
2496 At 400 keys, avg. speed is 2,578 keys/sec
2497 At 500 keys, avg. speed is 2,722 keys/sec
2498 At 600 keys, avg. speed is 2,628 keys/sec
2499 At 700 keys, avg. speed is 2,700 keys/sec
2500 At 800 keys, avg. speed is 2,607 keys/sec
2501 At 900 keys, avg. speed is 2,190 keys/sec
2502 At 1,000 keys, avg. speed is 2,570 keys/sec
2503 At 2,000 keys, avg. speed is 2,417 keys/sec
2504 At 3,000 keys, avg. speed is 1,982 keys/sec
2505 At 4,000 keys, avg. speed is 1,568 keys/sec
2506 At 5,000 keys, avg. speed is 1,533 keys/sec
2507 At 6,000 keys, avg. speed is 1,787 keys/sec
2508 At 7,000 keys, avg. speed is 1,977 keys/sec
2509 At 8,000 keys, avg. speed is 2,028 keys/sec
2510 At 9,000 keys, avg. speed is 2,077 keys/sec
2511 At 10,000 keys, avg. speed is 2,031 keys/sec
2512 At 20,000 keys, avg. speed is 1,970 keys/sec
2513 At 30,000 keys, avg. speed is 2,050 keys/sec
2514 At 40,000 keys, avg. speed is 2,073 keys/sec
2515 At 50,000 keys, avg. speed is 1,973 keys/sec
2516 At 60,000 keys, avg. speed is 1,914 keys/sec
2517 At 70,000 keys, avg. speed is 2,091 keys/sec
2518 At 80,000 keys, avg. speed is 2,103 keys/sec
2519 At 90,000 keys, avg. speed is 1,886 keys/sec
2520 At 100,000 keys, avg. speed is 1,970 keys/sec
2521 At 200,000 keys, avg. speed is 2,053 keys/sec
2522 At 300,000 keys, avg. speed is 1,697 keys/sec
2523 At 400,000 keys, avg. speed is 1,838 keys/sec
2524 At 500,000 keys, avg. speed is 1,941 keys/sec
2525 At 600,000 keys, avg. speed is 1,930 keys/sec
2526 At 700,000 keys, avg. speed is 1,735 keys/sec
2527 At 800,000 keys, avg. speed is 1,795 keys/sec
2528 At 900,000 keys, avg. speed is 1,221 keys/sec
2529 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2531 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2532 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2533 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2534 Run time was 12 min 3 sec.
2538 One of the great things about DBM::Deep is that it uses very little memory.
2539 Even with huge databases (1,000,000+ keys) you will not see much increased
2540 memory on your process. DBM::Deep relies solely on the filesystem for storing
2541 and fetching data. Here is output from I</usr/bin/top> before even opening a
2544 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2545 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2547 Basically the process is taking 2,716K of memory. And here is the same
2548 process after storing and fetching 1,000,000 keys:
2550 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2551 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2553 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2554 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2556 =head1 DB FILE FORMAT
2558 In case you were interested in the underlying DB file format, it is documented
2559 here in this section. You don't need to know this to use the module, it's just
2560 included for reference.
2564 DBM::Deep files always start with a 32-bit signature to identify the file type.
2565 This is at offset 0. The signature is "DPDB" in network byte order. This is
2566 checked when the file is opened.
2570 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2571 has a standard header containing the type of data, the length of data, and then
2572 the data itself. The type is a single character (1 byte), the length is a
2573 32-bit unsigned long in network byte order, and the data is, well, the data.
2574 Here is how it unfolds:
2578 Immediately after the 32-bit file signature is the I<Master Index> record.
2579 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2580 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2581 depending on how the DBM::Deep object was constructed.
2585 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2586 number). The first 8-bit char of the MD5 signature is the offset into the
2587 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2588 index element is a file offset of the next tag for the key/element in question,
2589 which is usually a I<Bucket List> tag (see below).
2593 The next tag I<could> be another index, depending on how many keys/elements
2594 exist. See L<RE-INDEXING> below for details.
2598 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2599 file offsets to where the actual data is stored. It starts with a standard
2600 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2601 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2602 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2603 When the list fills up, a I<Re-Index> operation is performed (See
2604 L<RE-INDEXING> below).
2608 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2609 index/value pair (in array mode). It starts with a standard tag header with
2610 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2611 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2612 header. The size reported in the tag header is only for the value, but then,
2613 just after the value is another size (32-bit unsigned long) and then the plain
2614 key itself. Since the value is likely to be fetched more often than the plain
2615 key, I figured it would be I<slightly> faster to store the value first.
2619 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2620 record for the nested structure, where the process begins all over again.
2624 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2625 exhausted. Then, when another key/element comes in, the list is converted to a
2626 new index record. However, this index will look at the next char in the MD5
2627 hash, and arrange new Bucket List pointers accordingly. This process is called
2628 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2629 17 (16 + new one) keys/elements are removed from the old Bucket List and
2630 inserted into the new index. Several new Bucket Lists are created in the
2631 process, as a new MD5 char from the key is being examined (it is unlikely that
2632 the keys will all share the same next char of their MD5s).
2636 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2637 when the Bucket Lists will turn into indexes, but the first round tends to
2638 happen right around 4,000 keys. You will see a I<slight> decrease in
2639 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2640 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2641 right around 900,000 keys. This process can continue nearly indefinitely --
2642 right up until the point the I<MD5> signatures start colliding with each other,
2643 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2644 getting struck by lightning while you are walking to cash in your tickets.
2645 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2646 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2647 this is 340 unodecillion, but don't quote me).
2651 When a new key/element is stored, the key (or index number) is first ran through
2652 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2653 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2654 for the first char of the signature (in this case I<b>). If it does not exist,
2655 a new I<Bucket List> is created for our key (and the next 15 future keys that
2656 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2657 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2658 this point, unless we are replacing an existing I<Bucket>), where the actual
2659 data will be stored.
2663 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2664 (or index number), then walking along the indexes. If there are enough
2665 keys/elements in this DB level, there might be nested indexes, each linked to
2666 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2667 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2668 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2669 plain key are stored.
2673 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2674 methods. In this process the indexes are walked systematically, and each key
2675 fetched in increasing MD5 order (which is why it appears random). Once the
2676 I<Bucket> is found, the value is skipped the plain key returned instead.
2677 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2678 alphabetically sorted. This only happens on an index-level -- as soon as the
2679 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2680 so it's pretty much undefined how the keys will come out -- just like Perl's
2683 =head1 CODE COVERAGE
2685 I use B<Devel::Cover> to test the code coverage of my tests, below is the B<Devel::Cover>
2686 report on this module's test suite.
2688 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2689 File stmt bran cond sub pod time total
2690 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2691 blib/lib/DBM/Deep.pm 93.7 82.5 71.9 96.5 25.9 82.8 87.9
2692 blib/lib/DBM/Deep/Array.pm 98.8 88.0 90.9 100.0 n/a 12.8 96.3
2693 blib/lib/DBM/Deep/Hash.pm 95.2 80.0 100.0 100.0 n/a 4.4 92.3
2694 Total 94.8 83.2 76.5 97.6 25.9 100.0 89.7
2695 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2699 Joseph Huckaby, L<jhuckaby@cpan.org>
2701 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2705 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2706 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2710 Copyright (c) 2002-2005 Joseph Huckaby. All Rights Reserved.
2711 This is free software, you may use it and distribute it under the
2712 same terms as Perl itself.