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};
287 $self->root->{fh} = undef;
292 # Given offset, signature and content, create tag and write to disk
294 my ($self, $offset, $sig, $content) = @_;
295 my $size = length($content);
299 seek($fh, $offset, SEEK_SET);
300 print($fh $sig . pack($DATA_LENGTH_PACK, $size) . $content );
302 if ($offset == $self->root->{end}) {
303 $self->root->{end} += SIG_SIZE + $DATA_LENGTH_SIZE + $size;
309 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
316 # Given offset, load single tag and return signature, size and data
323 seek($fh, $offset, SEEK_SET);
324 if (eof $fh) { return undef; }
327 read( $fh, $sig, SIG_SIZE);
330 read( $fh, $size, $DATA_LENGTH_SIZE);
331 $size = unpack($DATA_LENGTH_PACK, $size);
334 read( $fh, $buffer, $size);
339 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
346 # Given index tag, lookup single entry in index and return .
349 my ($tag, $index) = @_;
351 my $location = unpack($LONG_PACK, substr($tag->{content}, $index * $LONG_SIZE, $LONG_SIZE) );
352 if (!$location) { return; }
354 return $self->_load_tag( $location );
359 # Adds one key/value pair to bucket list, given offset, MD5 digest of key,
360 # plain (undigested) key and value.
363 my ($tag, $md5, $plain_key, $value) = @_;
364 my $keys = $tag->{content};
368 my $is_dbm_deep = eval { $value->isa( 'DBM::Deep' ) };
369 my $internal_ref = $is_dbm_deep && ($value->root eq $self->root);
374 # Iterate through buckets, seeing if this is a new entry or a replace.
376 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
377 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
378 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
381 # Found empty bucket (end of list). Populate and exit loop.
385 $location = $internal_ref
386 ? $value->base_offset
387 : $self->root->{end};
389 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
390 print($fh $md5 . pack($LONG_PACK, $location) );
393 elsif ($md5 eq $key) {
395 # Found existing bucket with same key. Replace with new value.
400 $location = $value->base_offset;
401 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
402 print($fh $md5 . pack($LONG_PACK, $location) );
405 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
407 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
410 # If value is a hash, array, or raw value with equal or less size, we can
411 # reuse the same content area of the database. Otherwise, we have to create
412 # a new content area at the EOF.
415 my $r = Scalar::Util::reftype( $value ) || '';
416 if ( $r eq 'HASH' || $r eq 'ARRAY' ) { $actual_length = $INDEX_SIZE; }
417 else { $actual_length = length($value); }
419 if ($actual_length <= $size) {
423 $location = $self->root->{end};
424 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE, SEEK_SET);
425 print($fh pack($LONG_PACK, $location) );
433 # If this is an internal reference, return now.
434 # No need to write value or plain key
441 # If bucket didn't fit into list, split into a new index level
444 seek($fh, $tag->{ref_loc}, SEEK_SET);
445 print($fh pack($LONG_PACK, $self->root->{end}) );
447 my $index_tag = $self->_create_tag($self->root->{end}, SIG_INDEX, chr(0) x $INDEX_SIZE);
450 $keys .= $md5 . pack($LONG_PACK, 0);
452 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
453 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
455 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
456 my $num = ord(substr($key, $tag->{ch} + 1, 1));
458 if ($offsets[$num]) {
459 my $offset = $offsets[$num] + SIG_SIZE + $DATA_LENGTH_SIZE;
460 seek($fh, $offset, SEEK_SET);
462 read( $fh, $subkeys, $BUCKET_LIST_SIZE);
464 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
465 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
467 seek($fh, $offset + ($k * $BUCKET_SIZE), SEEK_SET);
468 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
474 $offsets[$num] = $self->root->{end};
475 seek($fh, $index_tag->{offset} + ($num * $LONG_SIZE), SEEK_SET);
476 print($fh pack($LONG_PACK, $self->root->{end}) );
478 my $blist_tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
480 seek($fh, $blist_tag->{offset}, SEEK_SET);
481 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
486 $location ||= $self->root->{end};
487 } # re-index bucket list
490 # Seek to content area and store signature, value and plaintext key
494 seek($fh, $location, SEEK_SET);
497 # Write signature based on content type, set content length and write actual value.
499 my $r = Scalar::Util::reftype($value) || '';
501 print($fh TYPE_HASH );
502 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
503 $content_length = $INDEX_SIZE;
505 elsif ($r eq 'ARRAY') {
506 print($fh TYPE_ARRAY );
507 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
508 $content_length = $INDEX_SIZE;
510 elsif (!defined($value)) {
511 print($fh SIG_NULL );
512 print($fh pack($DATA_LENGTH_PACK, 0) );
516 print($fh SIG_DATA );
517 print($fh pack($DATA_LENGTH_PACK, length($value)) . $value );
518 $content_length = length($value);
522 # Plain key is stored AFTER value, as keys are typically fetched less often.
524 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
527 # If value is blessed, preserve class name
529 if ( $self->root->{autobless} ) {
530 my $value_class = Scalar::Util::blessed($value);
531 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
533 # Blessed ref -- will restore later
536 print($fh pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
537 $content_length += 1;
538 $content_length += $DATA_LENGTH_SIZE + length($value_class);
542 $content_length += 1;
547 # If this is a new content area, advance EOF counter
549 if ($location == $self->root->{end}) {
550 $self->root->{end} += SIG_SIZE;
551 $self->root->{end} += $DATA_LENGTH_SIZE + $content_length;
552 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
556 # If content is a hash or array, create new child DeepDB object and
557 # pass each key or element to it.
560 my $branch = DBM::Deep->new(
562 base_offset => $location,
565 foreach my $key (keys %{$value}) {
566 #$branch->{$key} = $value->{$key};
567 $branch->STORE( $key, $value->{$key} );
570 elsif ($r eq 'ARRAY') {
571 my $branch = DBM::Deep->new(
573 base_offset => $location,
577 foreach my $element (@{$value}) {
578 #$branch->[$index] = $element;
579 $branch->STORE( $index, $element );
587 return $self->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
590 sub _get_bucket_value {
592 # Fetch single value given tag and MD5 digested key.
595 my ($tag, $md5) = @_;
596 my $keys = $tag->{content};
601 # Iterate through buckets, looking for a key match
604 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
605 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
606 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
610 # Hit end of list, no match
615 if ( $md5 ne $key ) {
620 # Found match -- seek to offset and read signature
623 seek($fh, $subloc, SEEK_SET);
624 read( $fh, $signature, SIG_SIZE);
627 # If value is a hash or array, return new DeepDB object with correct offset
629 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
630 my $obj = DBM::Deep->new(
632 base_offset => $subloc,
636 if ($self->root->{autobless}) {
638 # Skip over value and plain key to see if object needs
641 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
644 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
645 if ($size) { seek($fh, $size, SEEK_CUR); }
648 read( $fh, $bless_bit, 1);
649 if (ord($bless_bit)) {
651 # Yes, object needs to be re-blessed
654 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
655 if ($size) { read( $fh, $class_name, $size); }
656 if ($class_name) { $obj = bless( $obj, $class_name ); }
664 # Otherwise return actual value
666 elsif ($signature eq SIG_DATA) {
669 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
670 if ($size) { read( $fh, $value, $size); }
675 # Key exists, but content is null
685 # Delete single key/value pair given tag and MD5 digested key.
688 my ($tag, $md5) = @_;
689 my $keys = $tag->{content};
694 # Iterate through buckets, looking for a key match
697 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
698 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
699 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
703 # Hit end of list, no match
708 if ( $md5 ne $key ) {
713 # Matched key -- delete bucket and return
715 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
716 print($fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
717 print($fh chr(0) x $BUCKET_SIZE );
727 # Check existence of single key given tag and MD5 digested key.
730 my ($tag, $md5) = @_;
731 my $keys = $tag->{content};
734 # Iterate through buckets, looking for a key match
737 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
738 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
739 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
743 # Hit end of list, no match
748 if ( $md5 ne $key ) {
753 # Matched key -- return true
761 sub _find_bucket_list {
763 # Locate offset for bucket list, given digested key
769 # Locate offset for bucket list using digest index system
772 my $tag = $self->_load_tag($self->base_offset);
773 if (!$tag) { return; }
775 while ($tag->{signature} ne SIG_BLIST) {
776 $tag = $self->_index_lookup($tag, ord(substr($md5, $ch, 1)));
777 if (!$tag) { return; }
784 sub _traverse_index {
786 # Scan index and recursively step into deeper levels, looking for next key.
788 my ($self, $offset, $ch, $force_return_next) = @_;
789 $force_return_next = undef unless $force_return_next;
791 my $tag = $self->_load_tag( $offset );
795 if ($tag->{signature} ne SIG_BLIST) {
796 my $content = $tag->{content};
798 if ($self->{return_next}) { $start = 0; }
799 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
801 for (my $index = $start; $index < 256; $index++) {
802 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
804 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
805 if (defined($result)) { return $result; }
809 $self->{return_next} = 1;
812 elsif ($tag->{signature} eq SIG_BLIST) {
813 my $keys = $tag->{content};
814 if ($force_return_next) { $self->{return_next} = 1; }
817 # Iterate through buckets, looking for a key match
819 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
820 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
821 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
825 # End of bucket list -- return to outer loop
827 $self->{return_next} = 1;
830 elsif ($key eq $self->{prev_md5}) {
832 # Located previous key -- return next one found
834 $self->{return_next} = 1;
837 elsif ($self->{return_next}) {
839 # Seek to bucket location and skip over signature
841 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
844 # Skip over value to get to plain key
847 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
848 if ($size) { seek($fh, $size, SEEK_CUR); }
851 # Read in plain key and return as scalar
854 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
855 if ($size) { read( $fh, $plain_key, $size); }
861 $self->{return_next} = 1;
862 } # tag is a bucket list
869 # Locate next key, given digested previous one
871 my $self = $_[0]->_get_self;
873 $self->{prev_md5} = $_[1] ? $_[1] : undef;
874 $self->{return_next} = 0;
877 # If the previous key was not specifed, start at the top and
878 # return the first one found.
880 if (!$self->{prev_md5}) {
881 $self->{prev_md5} = chr(0) x $HASH_SIZE;
882 $self->{return_next} = 1;
885 return $self->_traverse_index( $self->base_offset, 0 );
890 # If db locking is set, flock() the db file. If called multiple
891 # times before unlock(), then the same number of unlocks() must
892 # be called before the lock is released.
894 my $self = $_[0]->_get_self;
896 $type = LOCK_EX unless defined $type;
898 if (!defined($self->fh)) { return; }
900 if ($self->root->{locking}) {
901 if (!$self->root->{locked}) { flock($self->fh, $type); }
902 $self->root->{locked}++;
912 # If db locking is set, unlock the db file. See note in lock()
913 # regarding calling lock() multiple times.
915 my $self = $_[0]->_get_self;
917 if (!defined($self->fh)) { return; }
919 if ($self->root->{locking} && $self->root->{locked} > 0) {
920 $self->root->{locked}--;
921 if (!$self->root->{locked}) { flock($self->fh, LOCK_UN); }
929 #XXX These uses of ref() need verified
932 # Copy single level of keys or elements to new DB handle.
933 # Recurse for nested structures
935 my $self = $_[0]->_get_self;
938 if ($self->type eq TYPE_HASH) {
939 my $key = $self->first_key();
941 my $value = $self->get($key);
942 #XXX This doesn't work with autobless
943 if (!ref($value)) { $db_temp->{$key} = $value; }
945 my $type = $value->type;
946 if ($type eq TYPE_HASH) { $db_temp->{$key} = {}; }
947 else { $db_temp->{$key} = []; }
948 $value->_copy_node( $db_temp->{$key} );
950 $key = $self->next_key($key);
954 my $length = $self->length();
955 for (my $index = 0; $index < $length; $index++) {
956 my $value = $self->get($index);
957 if (!ref($value)) { $db_temp->[$index] = $value; }
958 #XXX NO tests for this code
960 my $type = $value->type;
961 if ($type eq TYPE_HASH) { $db_temp->[$index] = {}; }
962 else { $db_temp->[$index] = []; }
963 $value->_copy_node( $db_temp->[$index] );
971 # Recursively export into standard Perl hashes and arrays.
973 my $self = $_[0]->_get_self;
976 if ($self->type eq TYPE_HASH) { $temp = {}; }
977 elsif ($self->type eq TYPE_ARRAY) { $temp = []; }
980 $self->_copy_node( $temp );
988 # Recursively import Perl hash/array structure
990 #XXX This use of ref() seems to be ok
991 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
993 my $self = $_[0]->_get_self;
996 #XXX This use of ref() seems to be ok
999 # struct is not a reference, so just import based on our type
1003 if ($self->type eq TYPE_HASH) { $struct = {@_}; }
1004 elsif ($self->type eq TYPE_ARRAY) { $struct = [@_]; }
1007 my $r = Scalar::Util::reftype($struct) || '';
1008 if ($r eq "HASH" && $self->type eq TYPE_HASH) {
1009 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
1011 elsif ($r eq "ARRAY" && $self->type eq TYPE_ARRAY) {
1012 $self->push( @$struct );
1015 return $self->_throw_error("Cannot import: type mismatch");
1023 # Rebuild entire database into new file, then move
1024 # it back on top of original.
1026 my $self = $_[0]->_get_self;
1028 #XXX Need to create a new test for this
1029 # if ($self->root->{links} > 1) {
1030 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
1033 my $db_temp = DBM::Deep->new(
1034 file => $self->root->{file} . '.tmp',
1038 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
1042 $self->_copy_node( $db_temp );
1046 # Attempt to copy user, group and permissions over to new file
1048 my @stats = stat($self->fh);
1049 my $perms = $stats[2] & 07777;
1050 my $uid = $stats[4];
1051 my $gid = $stats[5];
1052 chown( $uid, $gid, $self->root->{file} . '.tmp' );
1053 chmod( $perms, $self->root->{file} . '.tmp' );
1055 # q.v. perlport for more information on this variable
1056 if ( $^O eq 'MSWin32' ) {
1058 # Potential race condition when optmizing on Win32 with locking.
1059 # The Windows filesystem requires that the filehandle be closed
1060 # before it is overwritten with rename(). This could be redone
1067 if (!rename $self->root->{file} . '.tmp', $self->root->{file}) {
1068 unlink $self->root->{file} . '.tmp';
1070 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
1082 # Make copy of object and return
1084 my $self = $_[0]->_get_self;
1086 return DBM::Deep->new(
1087 type => $self->type,
1088 base_offset => $self->base_offset,
1094 my %is_legal_filter = map {
1097 store_key store_value
1098 fetch_key fetch_value
1103 # Setup filter function for storing or fetching the key or value
1105 my $self = $_[0]->_get_self;
1106 my $type = lc $_[1];
1107 my $func = $_[2] ? $_[2] : undef;
1109 if ( $is_legal_filter{$type} ) {
1110 $self->root->{"filter_$type"} = $func;
1124 # Get access to the root structure
1126 my $self = $_[0]->_get_self;
1127 return $self->{root};
1132 # Get access to the raw FileHandle
1134 #XXX It will be useful, though, when we split out HASH and ARRAY
1135 my $self = $_[0]->_get_self;
1136 return $self->root->{fh};
1141 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
1143 my $self = $_[0]->_get_self;
1144 return $self->{type};
1149 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
1151 my $self = $_[0]->_get_self;
1152 return $self->{base_offset};
1157 # Get last error string, or undef if no error
1160 #? ( _get_self($_[0])->{root}->{error} or undef )
1161 ? ( $_[0]->_get_self->{root}->{error} or undef )
1171 # Store error string in self
1173 my $self = $_[0]->_get_self;
1174 my $error_text = $_[1];
1176 if ( Scalar::Util::blessed $self ) {
1177 $self->root->{error} = $error_text;
1179 unless ($self->root->{debug}) {
1180 die "DBM::Deep: $error_text\n";
1183 warn "DBM::Deep: $error_text\n";
1187 die "DBM::Deep: $error_text\n";
1195 my $self = $_[0]->_get_self;
1197 undef $self->root->{error};
1202 # Precalculate index, bucket and bucket list sizes
1205 #XXX I don't like this ...
1206 set_pack() unless defined $LONG_SIZE;
1208 $INDEX_SIZE = 256 * $LONG_SIZE;
1209 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
1210 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
1215 # Set pack/unpack modes (see file header for more)
1217 my ($long_s, $long_p, $data_s, $data_p) = @_;
1219 $LONG_SIZE = $long_s ? $long_s : 4;
1220 $LONG_PACK = $long_p ? $long_p : 'N';
1222 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
1223 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
1230 # Set key digest function (default is MD5)
1232 my ($digest_func, $hash_size) = @_;
1234 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
1235 $HASH_SIZE = $hash_size ? $hash_size : 16;
1241 # tie() methods (hashes and arrays)
1246 # Store single hash key/value or array element in database.
1248 my $self = $_[0]->_get_self;
1251 #XXX What is ref() checking here?
1252 #YYY User may be storing a hash, in which case we do not want it run
1253 #YYY through the filtering system
1254 my $value = ($self->root->{filter_store_value} && !ref($_[2]))
1255 ? $self->root->{filter_store_value}->($_[2])
1258 my $md5 = $DIGEST_FUNC->($key);
1261 # Make sure file is open
1263 if (!defined($self->fh) && !$self->_open()) {
1271 # Request exclusive lock for writing
1273 $self->lock( LOCK_EX );
1276 # If locking is enabled, set 'end' parameter again, in case another
1277 # DB instance appended to our file while we were unlocked.
1279 if ($self->root->{locking} || $self->root->{volatile}) {
1280 $self->root->{end} = (stat($fh))[7];
1284 # Locate offset for bucket list using digest index system
1286 my $tag = $self->_load_tag($self->base_offset);
1288 $tag = $self->_create_tag($self->base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1292 while ($tag->{signature} ne SIG_BLIST) {
1293 my $num = ord(substr($md5, $ch, 1));
1294 my $new_tag = $self->_index_lookup($tag, $num);
1296 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1297 seek($fh, $ref_loc, SEEK_SET);
1298 print($fh pack($LONG_PACK, $self->root->{end}) );
1300 $tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1301 $tag->{ref_loc} = $ref_loc;
1306 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1308 $tag->{ref_loc} = $ref_loc;
1315 # Add key/value to bucket list
1317 my $result = $self->_add_bucket( $tag, $md5, $key, $value );
1326 # Fetch single value or element given plain key or array index
1328 my $self = shift->_get_self;
1332 # Make sure file is open
1334 if (!defined($self->fh)) { $self->_open(); }
1336 my $md5 = $DIGEST_FUNC->($key);
1339 # Request shared lock for reading
1341 $self->lock( LOCK_SH );
1343 my $tag = $self->_find_bucket_list( $md5 );
1350 # Get value from bucket list
1352 my $result = $self->_get_bucket_value( $tag, $md5 );
1356 #XXX What is ref() checking here?
1357 return ($result && !ref($result) && $self->root->{filter_fetch_value})
1358 ? $self->root->{filter_fetch_value}->($result)
1364 # Delete single key/value pair or element given plain key or array index
1366 my $self = $_[0]->_get_self;
1369 my $unpacked_key = $key;
1370 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1371 my $md5 = $DIGEST_FUNC->($key);
1374 # Make sure file is open
1376 if (!defined($self->fh)) { $self->_open(); }
1379 # Request exclusive lock for writing
1381 $self->lock( LOCK_EX );
1383 my $tag = $self->_find_bucket_list( $md5 );
1392 my $value = $self->FETCH( $unpacked_key );
1393 my $result = $self->_delete_bucket( $tag, $md5 );
1396 # If this object is an array and the key deleted was on the end of the stack,
1397 # decrement the length variable.
1399 if ($result && ($self->type eq TYPE_ARRAY) && ($unpacked_key == $self->FETCHSIZE() - 1)) {
1400 $self->STORESIZE( $unpacked_key );
1410 # Check if a single key or element exists given plain key or array index
1412 my $self = $_[0]->_get_self;
1415 my $md5 = $DIGEST_FUNC->($key);
1418 # Make sure file is open
1420 if (!defined($self->fh)) { $self->_open(); }
1423 # Request shared lock for reading
1425 $self->lock( LOCK_SH );
1427 my $tag = $self->_find_bucket_list( $md5 );
1430 # For some reason, the built-in exists() function returns '' for false
1438 # Check if bucket exists and return 1 or ''
1440 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1449 # Clear all keys from hash, or all elements from array.
1451 my $self = $_[0]->_get_self;
1454 # Make sure file is open
1456 if (!defined($self->fh)) { $self->_open(); }
1459 # Request exclusive lock for writing
1461 $self->lock( LOCK_EX );
1465 seek($fh, $self->base_offset, SEEK_SET);
1471 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
1479 # Public method aliases
1481 sub put { (shift)->STORE( @_ ) }
1482 sub store { (shift)->STORE( @_ ) }
1483 sub get { (shift)->FETCH( @_ ) }
1484 sub fetch { (shift)->FETCH( @_ ) }
1485 sub delete { (shift)->DELETE( @_ ) }
1486 sub exists { (shift)->EXISTS( @_ ) }
1487 sub clear { (shift)->CLEAR( @_ ) }
1489 package DBM::Deep::_::Root;
1504 filter_store_key => undef,
1505 filter_store_value => undef,
1506 filter_fetch_key => undef,
1507 filter_fetch_value => undef,
1518 return unless $self;
1520 close $self->{fh} if $self->{fh};
1531 DBM::Deep - A pure perl multi-level hash/array DBM
1536 my $db = DBM::Deep->new( "foo.db" );
1538 $db->{key} = 'value'; # tie() style
1541 $db->put('key', 'value'); # OO style
1542 print $db->get('key');
1544 # true multi-level support
1545 $db->{my_complex} = [
1546 'hello', { perl => 'rules' },
1551 A unique flat-file database module, written in pure perl. True
1552 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1553 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1554 handle millions of keys and unlimited hash levels without significant
1555 slow-down. Written from the ground-up in pure perl -- this is NOT a
1556 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1557 Mac OS X and Windows.
1561 Hopefully you are using CPAN's excellent Perl module, which will download
1562 and install the module for you. If not, get the tarball, and run these
1574 Construction can be done OO-style (which is the recommended way), or using
1575 Perl's tie() function. Both are examined here.
1577 =head2 OO CONSTRUCTION
1579 The recommended way to construct a DBM::Deep object is to use the new()
1580 method, which gets you a blessed, tied hash or array reference.
1582 my $db = DBM::Deep->new( "foo.db" );
1584 This opens a new database handle, mapped to the file "foo.db". If this
1585 file does not exist, it will automatically be created. DB files are
1586 opened in "r+" (read/write) mode, and the type of object returned is a
1587 hash, unless otherwise specified (see L<OPTIONS> below).
1591 You can pass a number of options to the constructor to specify things like
1592 locking, autoflush, etc. This is done by passing an inline hash:
1594 my $db = DBM::Deep->new(
1600 Notice that the filename is now specified I<inside> the hash with
1601 the "file" parameter, as opposed to being the sole argument to the
1602 constructor. This is required if any options are specified.
1603 See L<OPTIONS> below for the complete list.
1607 You can also start with an array instead of a hash. For this, you must
1608 specify the C<type> parameter:
1610 my $db = DBM::Deep->new(
1612 type => DBM::Deep->TYPE_ARRAY
1615 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1616 a new DB file. If you create a DBM::Deep object with an existing file, the
1617 C<type> will be loaded from the file header, and ignored if it is passed
1620 =head2 TIE CONSTRUCTION
1622 Alternatively, you can create a DBM::Deep handle by using Perl's built-in
1623 tie() function. This is not ideal, because you get only a basic, tied hash
1624 (or array) which is not blessed, so you can't call any functions on it.
1627 tie %hash, "DBM::Deep", "foo.db";
1630 tie @array, "DBM::Deep", "bar.db";
1632 As with the OO constructor, you can replace the DB filename parameter with
1633 a hash containing one or more options (see L<OPTIONS> just below for the
1636 tie %hash, "DBM::Deep", {
1644 There are a number of options that can be passed in when constructing your
1645 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1651 Filename of the DB file to link the handle to. You can pass a full absolute
1652 filesystem path, partial path, or a plain filename if the file is in the
1653 current working directory. This is a required parameter.
1657 File open mode (read-only, read-write, etc.) string passed to Perl's FileHandle
1658 module. This is an optional parameter, and defaults to "r+" (read/write).
1659 B<Note:> If the default (r+) mode is selected, the file will also be auto-
1660 created if it doesn't exist.
1664 This parameter specifies what type of object to create, a hash or array. Use
1665 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1666 This only takes effect when beginning a new file. This is an optional
1667 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1671 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1672 function to lock the database in exclusive mode for writes, and shared mode for
1673 reads. Pass any true value to enable. This affects the base DB handle I<and
1674 any child hashes or arrays> that use the same DB file. This is an optional
1675 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1679 Specifies whether autoflush is to be enabled on the underlying FileHandle.
1680 This obviously slows down write operations, but is required if you may have
1681 multiple processes accessing the same DB file (also consider enable I<locking>
1682 or at least I<volatile>). Pass any true value to enable. This is an optional
1683 parameter, and defaults to 0 (disabled).
1687 If I<volatile> mode is enabled, DBM::Deep will stat() the DB file before each
1688 STORE() operation. This is required if an outside force may change the size of
1689 the file between transactions. Locking also implicitly enables volatile. This
1690 is useful if you want to use a different locking system or write your own. Pass
1691 any true value to enable. This is an optional parameter, and defaults to 0
1696 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1697 restore them when fetched. This is an B<experimental> feature, and does have
1698 side-effects. Basically, when hashes are re-blessed into their original
1699 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1700 able to call any DBM::Deep methods on them. You have been warned.
1701 This is an optional parameter, and defaults to 0 (disabled).
1705 See L<FILTERS> below.
1709 Setting I<debug> mode will make all errors non-fatal, dump them out to
1710 STDERR, and continue on. This is for debugging purposes only, and probably
1711 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1715 Instead of passing a file path, you can instead pass a handle to an pre-opened
1716 filehandle. Note: Beware of using the magick *DATA handle, as this actually
1717 contains your entire Perl script, as well as the data following the __DATA__
1718 marker. This will not work, because DBM::Deep uses absolute seek()s into the
1719 file. Instead, consider reading *DATA into an IO::Scalar handle, then passing
1724 =head1 TIE INTERFACE
1726 With DBM::Deep you can access your databases using Perl's standard hash/array
1727 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can treat
1728 them as such. DBM::Deep will intercept all reads/writes and direct them to the right
1729 place -- the DB file. This has nothing to do with the L<TIE CONSTRUCTION>
1730 section above. This simply tells you how to use DBM::Deep using regular hashes
1731 and arrays, rather than calling functions like C<get()> and C<put()> (although those
1732 work too). It is entirely up to you how to want to access your databases.
1736 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1737 or even nested hashes (or arrays) using standard Perl syntax:
1739 my $db = DBM::Deep->new( "foo.db" );
1741 $db->{mykey} = "myvalue";
1743 $db->{myhash}->{subkey} = "subvalue";
1745 print $db->{myhash}->{subkey} . "\n";
1747 You can even step through hash keys using the normal Perl C<keys()> function:
1749 foreach my $key (keys %$db) {
1750 print "$key: " . $db->{$key} . "\n";
1753 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1754 pushes them onto an array, all before the loop even begins. If you have an
1755 extra large hash, this may exhaust Perl's memory. Instead, consider using
1756 Perl's C<each()> function, which pulls keys/values one at a time, using very
1759 while (my ($key, $value) = each %$db) {
1760 print "$key: $value\n";
1763 Please note that when using C<each()>, you should always pass a direct
1764 hash reference, not a lookup. Meaning, you should B<never> do this:
1767 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1769 This causes an infinite loop, because for each iteration, Perl is calling
1770 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1771 it effectively keeps returning the first key over and over again. Instead,
1772 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1776 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1777 reference. This includes inserting, removing and manipulating elements,
1778 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1779 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1780 or simply be a nested array reference inside a hash. Example:
1782 my $db = DBM::Deep->new(
1783 file => "foo-array.db",
1784 type => DBM::Deep->TYPE_ARRAY
1788 push @$db, "bar", "baz";
1789 unshift @$db, "bah";
1791 my $last_elem = pop @$db; # baz
1792 my $first_elem = shift @$db; # bah
1793 my $second_elem = $db->[1]; # bar
1795 my $num_elements = scalar @$db;
1799 In addition to the I<tie()> interface, you can also use a standard OO interface
1800 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1801 array) has its own methods, but both types share the following common methods:
1802 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1808 Stores a new hash key/value pair, or sets an array element value. Takes two
1809 arguments, the hash key or array index, and the new value. The value can be
1810 a scalar, hash ref or array ref. Returns true on success, false on failure.
1812 $db->put("foo", "bar"); # for hashes
1813 $db->put(1, "bar"); # for arrays
1817 Fetches the value of a hash key or array element. Takes one argument: the hash
1818 key or array index. Returns a scalar, hash ref or array ref, depending on the
1821 my $value = $db->get("foo"); # for hashes
1822 my $value = $db->get(1); # for arrays
1826 Checks if a hash key or array index exists. Takes one argument: the hash key
1827 or array index. Returns true if it exists, false if not.
1829 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1830 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1834 Deletes one hash key/value pair or array element. Takes one argument: the hash
1835 key or array index. Returns true on success, false if not found. For arrays,
1836 the remaining elements located after the deleted element are NOT moved over.
1837 The deleted element is essentially just undefined, which is exactly how Perl's
1838 internal arrays work. Please note that the space occupied by the deleted
1839 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1840 below for details and workarounds.
1842 $db->delete("foo"); # for hashes
1843 $db->delete(1); # for arrays
1847 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1848 value. Please note that the space occupied by the deleted keys/values or
1849 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1850 details and workarounds.
1852 $db->clear(); # hashes or arrays
1858 For hashes, DBM::Deep supports all the common methods described above, and the
1859 following additional methods: C<first_key()> and C<next_key()>.
1865 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1866 fetched in an undefined order (which appears random). Takes no arguments,
1867 returns the key as a scalar value.
1869 my $key = $db->first_key();
1873 Returns the "next" key in the hash, given the previous one as the sole argument.
1874 Returns undef if there are no more keys to be fetched.
1876 $key = $db->next_key($key);
1880 Here are some examples of using hashes:
1882 my $db = DBM::Deep->new( "foo.db" );
1884 $db->put("foo", "bar");
1885 print "foo: " . $db->get("foo") . "\n";
1887 $db->put("baz", {}); # new child hash ref
1888 $db->get("baz")->put("buz", "biz");
1889 print "buz: " . $db->get("baz")->get("buz") . "\n";
1891 my $key = $db->first_key();
1893 print "$key: " . $db->get($key) . "\n";
1894 $key = $db->next_key($key);
1897 if ($db->exists("foo")) { $db->delete("foo"); }
1901 For arrays, DBM::Deep supports all the common methods described above, and the
1902 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1903 C<unshift()> and C<splice()>.
1909 Returns the number of elements in the array. Takes no arguments.
1911 my $len = $db->length();
1915 Adds one or more elements onto the end of the array. Accepts scalars, hash
1916 refs or array refs. No return value.
1918 $db->push("foo", "bar", {});
1922 Fetches the last element in the array, and deletes it. Takes no arguments.
1923 Returns undef if array is empty. Returns the element value.
1925 my $elem = $db->pop();
1929 Fetches the first element in the array, deletes it, then shifts all the
1930 remaining elements over to take up the space. Returns the element value. This
1931 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1934 my $elem = $db->shift();
1938 Inserts one or more elements onto the beginning of the array, shifting all
1939 existing elements over to make room. Accepts scalars, hash refs or array refs.
1940 No return value. This method is not recommended with large arrays -- see
1941 <LARGE ARRAYS> below for details.
1943 $db->unshift("foo", "bar", {});
1947 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1948 -f splice> for usage -- it is too complicated to document here. This method is
1949 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1953 Here are some examples of using arrays:
1955 my $db = DBM::Deep->new(
1957 type => DBM::Deep->TYPE_ARRAY
1960 $db->push("bar", "baz");
1961 $db->unshift("foo");
1964 my $len = $db->length();
1965 print "length: $len\n"; # 4
1967 for (my $k=0; $k<$len; $k++) {
1968 print "$k: " . $db->get($k) . "\n";
1971 $db->splice(1, 2, "biz", "baf");
1973 while (my $elem = shift @$db) {
1974 print "shifted: $elem\n";
1979 Enable automatic file locking by passing a true value to the C<locking>
1980 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1982 my $db = DBM::Deep->new(
1987 This causes DBM::Deep to C<flock()> the underlying FileHandle object with exclusive
1988 mode for writes, and shared mode for reads. This is required if you have
1989 multiple processes accessing the same database file, to avoid file corruption.
1990 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1991 NFS> below for more.
1993 =head2 EXPLICIT LOCKING
1995 You can explicitly lock a database, so it remains locked for multiple
1996 transactions. This is done by calling the C<lock()> method, and passing an
1997 optional lock mode argument (defaults to exclusive mode). This is particularly
1998 useful for things like counters, where the current value needs to be fetched,
1999 then incremented, then stored again.
2002 my $counter = $db->get("counter");
2004 $db->put("counter", $counter);
2013 You can pass C<lock()> an optional argument, which specifies which mode to use
2014 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2015 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2016 same as the constants defined in Perl's C<Fcntl> module.
2018 $db->lock( DBM::Deep->LOCK_SH );
2022 If you want to implement your own file locking scheme, be sure to create your
2023 DBM::Deep objects setting the C<volatile> option to true. This hints to DBM::Deep
2024 that the DB file may change between transactions. See L<LOW-LEVEL ACCESS>
2027 =head1 IMPORTING/EXPORTING
2029 You can import existing complex structures by calling the C<import()> method,
2030 and export an entire database into an in-memory structure using the C<export()>
2031 method. Both are examined here.
2035 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2036 walking the structure and adding keys/elements to the database as you go,
2037 simply pass a reference to the C<import()> method. This recursively adds
2038 everything to an existing DBM::Deep object for you. Here is an example:
2043 array1 => [ "elem0", "elem1", "elem2" ],
2045 subkey1 => "subvalue1",
2046 subkey2 => "subvalue2"
2050 my $db = DBM::Deep->new( "foo.db" );
2051 $db->import( $struct );
2053 print $db->{key1} . "\n"; # prints "value1"
2055 This recursively imports the entire C<$struct> object into C<$db>, including
2056 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2057 keys are merged with the existing ones, replacing if they already exist.
2058 The C<import()> method can be called on any database level (not just the base
2059 level), and works with both hash and array DB types.
2063 B<Note:> Make sure your existing structure has no circular references in it.
2064 These will cause an infinite loop when importing.
2068 Calling the C<export()> method on an existing DBM::Deep object will return
2069 a reference to a new in-memory copy of the database. The export is done
2070 recursively, so all nested hashes/arrays are all exported to standard Perl
2071 objects. Here is an example:
2073 my $db = DBM::Deep->new( "foo.db" );
2075 $db->{key1} = "value1";
2076 $db->{key2} = "value2";
2078 $db->{hash1}->{subkey1} = "subvalue1";
2079 $db->{hash1}->{subkey2} = "subvalue2";
2081 my $struct = $db->export();
2083 print $struct->{key1} . "\n"; # prints "value1"
2085 This makes a complete copy of the database in memory, and returns a reference
2086 to it. The C<export()> method can be called on any database level (not just
2087 the base level), and works with both hash and array DB types. Be careful of
2088 large databases -- you can store a lot more data in a DBM::Deep object than an
2089 in-memory Perl structure.
2093 B<Note:> Make sure your database has no circular references in it.
2094 These will cause an infinite loop when exporting.
2098 DBM::Deep has a number of hooks where you can specify your own Perl function
2099 to perform filtering on incoming or outgoing data. This is a perfect
2100 way to extend the engine, and implement things like real-time compression or
2101 encryption. Filtering applies to the base DB level, and all child hashes /
2102 arrays. Filter hooks can be specified when your DBM::Deep object is first
2103 constructed, or by calling the C<set_filter()> method at any time. There are
2104 four available filter hooks, described below:
2108 =item * filter_store_key
2110 This filter is called whenever a hash key is stored. It
2111 is passed the incoming key, and expected to return a transformed key.
2113 =item * filter_store_value
2115 This filter is called whenever a hash key or array element is stored. It
2116 is passed the incoming value, and expected to return a transformed value.
2118 =item * filter_fetch_key
2120 This filter is called whenever a hash key is fetched (i.e. via
2121 C<first_key()> or C<next_key()>). It is passed the transformed key,
2122 and expected to return the plain key.
2124 =item * filter_fetch_value
2126 This filter is called whenever a hash key or array element is fetched.
2127 It is passed the transformed value, and expected to return the plain value.
2131 Here are the two ways to setup a filter hook:
2133 my $db = DBM::Deep->new(
2135 filter_store_value => \&my_filter_store,
2136 filter_fetch_value => \&my_filter_fetch
2141 $db->set_filter( "filter_store_value", \&my_filter_store );
2142 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2144 Your filter function will be called only when dealing with SCALAR keys or
2145 values. When nested hashes and arrays are being stored/fetched, filtering
2146 is bypassed. Filters are called as static functions, passed a single SCALAR
2147 argument, and expected to return a single SCALAR value. If you want to
2148 remove a filter, set the function reference to C<undef>:
2150 $db->set_filter( "filter_store_value", undef );
2152 =head2 REAL-TIME ENCRYPTION EXAMPLE
2154 Here is a working example that uses the I<Crypt::Blowfish> module to
2155 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2156 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2157 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2160 use Crypt::Blowfish;
2163 my $cipher = Crypt::CBC->new({
2164 'key' => 'my secret key',
2165 'cipher' => 'Blowfish',
2167 'regenerate_key' => 0,
2168 'padding' => 'space',
2172 my $db = DBM::Deep->new(
2173 file => "foo-encrypt.db",
2174 filter_store_key => \&my_encrypt,
2175 filter_store_value => \&my_encrypt,
2176 filter_fetch_key => \&my_decrypt,
2177 filter_fetch_value => \&my_decrypt,
2180 $db->{key1} = "value1";
2181 $db->{key2} = "value2";
2182 print "key1: " . $db->{key1} . "\n";
2183 print "key2: " . $db->{key2} . "\n";
2189 return $cipher->encrypt( $_[0] );
2192 return $cipher->decrypt( $_[0] );
2195 =head2 REAL-TIME COMPRESSION EXAMPLE
2197 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2198 compression / decompression of keys & values with DBM::Deep Filters.
2199 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2200 more on I<Compress::Zlib>.
2205 my $db = DBM::Deep->new(
2206 file => "foo-compress.db",
2207 filter_store_key => \&my_compress,
2208 filter_store_value => \&my_compress,
2209 filter_fetch_key => \&my_decompress,
2210 filter_fetch_value => \&my_decompress,
2213 $db->{key1} = "value1";
2214 $db->{key2} = "value2";
2215 print "key1: " . $db->{key1} . "\n";
2216 print "key2: " . $db->{key2} . "\n";
2222 return Compress::Zlib::memGzip( $_[0] ) ;
2225 return Compress::Zlib::memGunzip( $_[0] ) ;
2228 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2229 actually numerical index numbers, and are not filtered.
2231 =head1 ERROR HANDLING
2233 Most DBM::Deep methods return a true value for success, and call die() on
2234 failure. You can wrap calls in an eval block to catch the die. Also, the
2235 actual error message is stored in an internal scalar, which can be fetched by
2236 calling the C<error()> method.
2238 my $db = DBM::Deep->new( "foo.db" ); # create hash
2239 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2241 print $db->error(); # prints error message
2243 You can then call C<clear_error()> to clear the current error state.
2247 If you set the C<debug> option to true when creating your DBM::Deep object,
2248 all errors are considered NON-FATAL, and dumped to STDERR. This is only
2249 for debugging purposes.
2251 =head1 LARGEFILE SUPPORT
2253 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2254 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2255 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2256 by calling the static C<set_pack()> method before you do anything else.
2258 DBM::Deep::set_pack(8, 'Q');
2260 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2261 instead of 32-bit longs. After setting these values your DB files have a
2262 theoretical maximum size of 16 XB (exabytes).
2266 B<Note:> Changing these values will B<NOT> work for existing database files.
2267 Only change this for new files, and make sure it stays set consistently
2268 throughout the file's life. If you do set these values, you can no longer
2269 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2270 back to 32-bit mode.
2274 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2275 only a 32-bit Perl. However, I have received user reports that this does
2278 =head1 LOW-LEVEL ACCESS
2280 If you require low-level access to the underlying FileHandle that DBM::Deep uses,
2281 you can call the C<fh()> method, which returns the handle:
2285 This method can be called on the root level of the datbase, or any child
2286 hashes or arrays. All levels share a I<root> structure, which contains things
2287 like the FileHandle, a reference counter, and all your options you specified
2288 when you created the object. You can get access to this root structure by
2289 calling the C<root()> method.
2291 my $root = $db->root();
2293 This is useful for changing options after the object has already been created,
2294 such as enabling/disabling locking, volatile or debug modes. You can also
2295 store your own temporary user data in this structure (be wary of name
2296 collision), which is then accessible from any child hash or array.
2298 =head1 CUSTOM DIGEST ALGORITHM
2300 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2301 keys. However you can override this, and use another algorithm (such as SHA-256)
2302 or even write your own. But please note that DBM::Deep currently expects zero
2303 collisions, so your algorithm has to be I<perfect>, so to speak.
2304 Collision detection may be introduced in a later version.
2308 You can specify a custom digest algorithm by calling the static C<set_digest()>
2309 function, passing a reference to a subroutine, and the length of the algorithm's
2310 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2311 objects. Here is a working example that uses a 256-bit hash from the
2312 I<Digest::SHA256> module. Please see
2313 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2318 my $context = Digest::SHA256::new(256);
2320 DBM::Deep::set_digest( \&my_digest, 32 );
2322 my $db = DBM::Deep->new( "foo-sha.db" );
2324 $db->{key1} = "value1";
2325 $db->{key2} = "value2";
2326 print "key1: " . $db->{key1} . "\n";
2327 print "key2: " . $db->{key2} . "\n";
2333 return substr( $context->hash($_[0]), 0, 32 );
2336 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2337 of bytes you specify in the C<set_digest()> function (in this case 32).
2339 =head1 CIRCULAR REFERENCES
2341 DBM::Deep has B<experimental> support for circular references. Meaning you
2342 can have a nested hash key or array element that points to a parent object.
2343 This relationship is stored in the DB file, and is preserved between sessions.
2346 my $db = DBM::Deep->new( "foo.db" );
2349 $db->{circle} = $db; # ref to self
2351 print $db->{foo} . "\n"; # prints "foo"
2352 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2354 One catch is, passing the object to a function that recursively walks the
2355 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2356 C<export()> methods) will result in an infinite loop. The other catch is,
2357 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2358 or C<next_key()> methods), you will get the I<target object's key>, not the
2359 ref's key. This gets even more interesting with the above example, where
2360 the I<circle> key points to the base DB object, which technically doesn't
2361 have a key. So I made DBM::Deep return "[base]" as the key name in that
2364 =head1 CAVEATS / ISSUES / BUGS
2366 This section describes all the known issues with DBM::Deep. It you have found
2367 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2369 =head2 UNUSED SPACE RECOVERY
2371 One major caveat with DBM::Deep is that space occupied by existing keys and
2372 values is not recovered when they are deleted. Meaning if you keep deleting
2373 and adding new keys, your file will continuously grow. I am working on this,
2374 but in the meantime you can call the built-in C<optimize()> method from time to
2375 time (perhaps in a crontab or something) to recover all your unused space.
2377 $db->optimize(); # returns true on success
2379 This rebuilds the ENTIRE database into a new file, then moves it on top of
2380 the original. The new file will have no unused space, thus it will take up as
2381 little disk space as possible. Please note that this operation can take
2382 a long time for large files, and you need enough disk space to temporarily hold
2383 2 copies of your DB file. The temporary file is created in the same directory
2384 as the original, named with a ".tmp" extension, and is deleted when the
2385 operation completes. Oh, and if locking is enabled, the DB is automatically
2386 locked for the entire duration of the copy.
2390 B<WARNING:> Only call optimize() on the top-level node of the database, and
2391 make sure there are no child references lying around. DBM::Deep keeps a reference
2392 counter, and if it is greater than 1, optimize() will abort and return undef.
2394 =head2 AUTOVIVIFICATION
2396 Unfortunately, autovivification doesn't work with tied hashes. This appears to
2397 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
2398 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
2399 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
2400 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
2403 $db->{foo}->{bar} = "hello";
2405 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
2406 being an empty hash. Try this instead, which works fine:
2408 $db->{foo} = { bar => "hello" };
2410 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
2411 the execution path, and Perl indeed passes an empty hash to the STORE() method.
2412 Probably a bug in Perl.
2414 =head2 FILE CORRUPTION
2416 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2417 for a 32-bit signature when opened, but other corruption in files can cause
2418 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2419 stuck in an infinite loop depending on the level of corruption. File write
2420 operations are not checked for failure (for speed), so if you happen to run
2421 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2422 be addressed in a later version of DBM::Deep.
2426 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2427 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2428 about setting up your NFS server with a locking daemon, then using lockf() to
2429 lock your files, but your milage may vary there as well. From what I
2430 understand, there is no real way to do it. However, if you need access to the
2431 underlying FileHandle in DBM::Deep for using some other kind of locking scheme like
2432 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2434 =head2 COPYING OBJECTS
2436 Beware of copying tied objects in Perl. Very strange things can happen.
2437 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2438 returns a new, blessed, tied hash or array to the same level in the DB.
2440 my $copy = $db->clone();
2444 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2445 These functions cause every element in the array to move, which can be murder
2446 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2447 a different location. This may be addressed in a later version.
2451 This section discusses DBM::Deep's speed and memory usage.
2455 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2456 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2457 multi-level hash/array support, and cross-platform FTPable files. Even so,
2458 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2459 with huge databases. Here is some test data:
2461 Adding 1,000,000 keys to new DB file...
2463 At 100 keys, avg. speed is 2,703 keys/sec
2464 At 200 keys, avg. speed is 2,642 keys/sec
2465 At 300 keys, avg. speed is 2,598 keys/sec
2466 At 400 keys, avg. speed is 2,578 keys/sec
2467 At 500 keys, avg. speed is 2,722 keys/sec
2468 At 600 keys, avg. speed is 2,628 keys/sec
2469 At 700 keys, avg. speed is 2,700 keys/sec
2470 At 800 keys, avg. speed is 2,607 keys/sec
2471 At 900 keys, avg. speed is 2,190 keys/sec
2472 At 1,000 keys, avg. speed is 2,570 keys/sec
2473 At 2,000 keys, avg. speed is 2,417 keys/sec
2474 At 3,000 keys, avg. speed is 1,982 keys/sec
2475 At 4,000 keys, avg. speed is 1,568 keys/sec
2476 At 5,000 keys, avg. speed is 1,533 keys/sec
2477 At 6,000 keys, avg. speed is 1,787 keys/sec
2478 At 7,000 keys, avg. speed is 1,977 keys/sec
2479 At 8,000 keys, avg. speed is 2,028 keys/sec
2480 At 9,000 keys, avg. speed is 2,077 keys/sec
2481 At 10,000 keys, avg. speed is 2,031 keys/sec
2482 At 20,000 keys, avg. speed is 1,970 keys/sec
2483 At 30,000 keys, avg. speed is 2,050 keys/sec
2484 At 40,000 keys, avg. speed is 2,073 keys/sec
2485 At 50,000 keys, avg. speed is 1,973 keys/sec
2486 At 60,000 keys, avg. speed is 1,914 keys/sec
2487 At 70,000 keys, avg. speed is 2,091 keys/sec
2488 At 80,000 keys, avg. speed is 2,103 keys/sec
2489 At 90,000 keys, avg. speed is 1,886 keys/sec
2490 At 100,000 keys, avg. speed is 1,970 keys/sec
2491 At 200,000 keys, avg. speed is 2,053 keys/sec
2492 At 300,000 keys, avg. speed is 1,697 keys/sec
2493 At 400,000 keys, avg. speed is 1,838 keys/sec
2494 At 500,000 keys, avg. speed is 1,941 keys/sec
2495 At 600,000 keys, avg. speed is 1,930 keys/sec
2496 At 700,000 keys, avg. speed is 1,735 keys/sec
2497 At 800,000 keys, avg. speed is 1,795 keys/sec
2498 At 900,000 keys, avg. speed is 1,221 keys/sec
2499 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2501 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2502 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2503 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2504 Run time was 12 min 3 sec.
2508 One of the great things about DBM::Deep is that it uses very little memory.
2509 Even with huge databases (1,000,000+ keys) you will not see much increased
2510 memory on your process. DBM::Deep relies solely on the filesystem for storing
2511 and fetching data. Here is output from I</usr/bin/top> before even opening a
2514 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2515 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2517 Basically the process is taking 2,716K of memory. And here is the same
2518 process after storing and fetching 1,000,000 keys:
2520 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2521 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2523 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2524 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2526 =head1 DB FILE FORMAT
2528 In case you were interested in the underlying DB file format, it is documented
2529 here in this section. You don't need to know this to use the module, it's just
2530 included for reference.
2534 DBM::Deep files always start with a 32-bit signature to identify the file type.
2535 This is at offset 0. The signature is "DPDB" in network byte order. This is
2536 checked when the file is opened.
2540 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2541 has a standard header containing the type of data, the length of data, and then
2542 the data itself. The type is a single character (1 byte), the length is a
2543 32-bit unsigned long in network byte order, and the data is, well, the data.
2544 Here is how it unfolds:
2548 Immediately after the 32-bit file signature is the I<Master Index> record.
2549 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2550 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2551 depending on how the DBM::Deep object was constructed.
2555 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2556 number). The first 8-bit char of the MD5 signature is the offset into the
2557 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2558 index element is a file offset of the next tag for the key/element in question,
2559 which is usually a I<Bucket List> tag (see below).
2563 The next tag I<could> be another index, depending on how many keys/elements
2564 exist. See L<RE-INDEXING> below for details.
2568 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2569 file offsets to where the actual data is stored. It starts with a standard
2570 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2571 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2572 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2573 When the list fills up, a I<Re-Index> operation is performed (See
2574 L<RE-INDEXING> below).
2578 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2579 index/value pair (in array mode). It starts with a standard tag header with
2580 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2581 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2582 header. The size reported in the tag header is only for the value, but then,
2583 just after the value is another size (32-bit unsigned long) and then the plain
2584 key itself. Since the value is likely to be fetched more often than the plain
2585 key, I figured it would be I<slightly> faster to store the value first.
2589 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2590 record for the nested structure, where the process begins all over again.
2594 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2595 exhausted. Then, when another key/element comes in, the list is converted to a
2596 new index record. However, this index will look at the next char in the MD5
2597 hash, and arrange new Bucket List pointers accordingly. This process is called
2598 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2599 17 (16 + new one) keys/elements are removed from the old Bucket List and
2600 inserted into the new index. Several new Bucket Lists are created in the
2601 process, as a new MD5 char from the key is being examined (it is unlikely that
2602 the keys will all share the same next char of their MD5s).
2606 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2607 when the Bucket Lists will turn into indexes, but the first round tends to
2608 happen right around 4,000 keys. You will see a I<slight> decrease in
2609 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2610 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2611 right around 900,000 keys. This process can continue nearly indefinitely --
2612 right up until the point the I<MD5> signatures start colliding with each other,
2613 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2614 getting struck by lightning while you are walking to cash in your tickets.
2615 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2616 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2617 this is 340 unodecillion, but don't quote me).
2621 When a new key/element is stored, the key (or index number) is first ran through
2622 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2623 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2624 for the first char of the signature (in this case I<b>). If it does not exist,
2625 a new I<Bucket List> is created for our key (and the next 15 future keys that
2626 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2627 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2628 this point, unless we are replacing an existing I<Bucket>), where the actual
2629 data will be stored.
2633 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2634 (or index number), then walking along the indexes. If there are enough
2635 keys/elements in this DB level, there might be nested indexes, each linked to
2636 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2637 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2638 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2639 plain key are stored.
2643 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2644 methods. In this process the indexes are walked systematically, and each key
2645 fetched in increasing MD5 order (which is why it appears random). Once the
2646 I<Bucket> is found, the value is skipped the plain key returned instead.
2647 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2648 alphabetically sorted. This only happens on an index-level -- as soon as the
2649 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2650 so it's pretty much undefined how the keys will come out -- just like Perl's
2653 =head1 CODE COVERAGE
2655 I use B<Devel::Cover> to test the code coverage of my tests, below is the B<Devel::Cover> report on this
2656 module's test suite.
2658 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2659 File stmt bran cond sub pod time total
2660 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2661 blib/lib/DBM/Deep.pm 93.9 82.4 74.7 97.9 10.5 85.7 88.0
2662 blib/lib/DBM/Deep/Array.pm 97.8 84.6 50.0 100.0 n/a 9.0 94.6
2663 blib/lib/DBM/Deep/Hash.pm 93.9 87.5 100.0 100.0 n/a 5.3 93.4
2664 Total 94.4 82.9 75.8 98.5 10.5 100.0 89.0
2665 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2669 Joseph Huckaby, L<jhuckaby@cpan.org>
2671 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2675 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2676 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2680 Copyright (c) 2002-2005 Joseph Huckaby. All Rights Reserved.
2681 This is free software, you may use it and distribute it under the
2682 same terms as Perl itself.