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 # locking implicitly enables autoflush
176 if ($args->{locking}) { $args->{autoflush} = 1; }
178 $self->{root} = exists $args->{root}
180 : DBM::Deep::_::Root->new( $args );
182 if (!defined($self->fh)) { $self->_open(); }
189 require DBM::Deep::Hash;
190 return DBM::Deep::Hash->TIEHASH( @_ );
195 require DBM::Deep::Array;
196 return DBM::Deep::Array->TIEARRAY( @_ );
199 #XXX Unneeded now ...
205 # Open a fh to the database, create if nonexistent.
206 # Make sure file signature matches DBM::Deep spec.
208 my $self = $_[0]->_get_self;
210 if (defined($self->fh)) { $self->_close(); }
213 # Theoretically, adding O_BINARY should remove the need for the binmode
214 # Of course, testing it is going to be ... interesting.
215 my $flags = O_RDWR | O_CREAT | O_BINARY;
218 sysopen( $fh, $self->root->{file}, $flags )
220 $self->root->{fh} = $fh;
221 }; if ($@ ) { $self->_throw_error( "Received error: $@\n" ); }
222 if (! defined($self->fh)) {
223 return $self->_throw_error("Cannot sysopen file: " . $self->root->{file} . ": $!");
228 #XXX Can we remove this by using the right sysopen() flags?
229 # Maybe ... q.v. above
230 binmode $fh; # for win32
232 if ($self->root->{autoflush}) {
233 my $old = select $fh;
239 seek($fh, 0, SEEK_SET);
242 my $bytes_read = read( $fh, $signature, length(SIG_FILE));
245 # File is empty -- write signature and master index
248 seek($fh, 0, SEEK_SET);
250 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
252 my $plain_key = "[base]";
253 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
255 # Flush the filehandle
256 my $old_fh = select $fh;
262 my @stats = stat($fh);
263 $self->root->{inode} = $stats[1];
264 $self->root->{end} = $stats[7];
270 # Check signature was valid
272 unless ($signature eq SIG_FILE) {
274 return $self->_throw_error("Signature not found -- file is not a Deep DB");
277 my @stats = stat($fh);
278 $self->root->{inode} = $stats[1];
279 $self->root->{end} = $stats[7];
282 # Get our type from master index signature
284 my $tag = $self->_load_tag($self->base_offset);
286 #XXX We probably also want to store the hash algorithm name and not assume anything
287 #XXX The cool thing would be to allow a different hashing algorithm at every level
290 return $self->_throw_error("Corrupted file, no master index record");
292 if ($self->{type} ne $tag->{signature}) {
293 return $self->_throw_error("File type mismatch");
303 my $self = $_[0]->_get_self;
304 close $self->root->{fh} if $self->root->{fh};
305 $self->root->{fh} = undef;
310 # Given offset, signature and content, create tag and write to disk
312 my ($self, $offset, $sig, $content) = @_;
313 my $size = length($content);
317 seek($fh, $offset, SEEK_SET);
318 print($fh $sig . pack($DATA_LENGTH_PACK, $size) . $content );
320 if ($offset == $self->root->{end}) {
321 $self->root->{end} += SIG_SIZE + $DATA_LENGTH_SIZE + $size;
327 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
334 # Given offset, load single tag and return signature, size and data
341 seek($fh, $offset, SEEK_SET);
342 if (eof $fh) { return undef; }
345 read( $fh, $sig, SIG_SIZE);
348 read( $fh, $size, $DATA_LENGTH_SIZE);
349 $size = unpack($DATA_LENGTH_PACK, $size);
352 read( $fh, $buffer, $size);
357 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
364 # Given index tag, lookup single entry in index and return .
367 my ($tag, $index) = @_;
369 my $location = unpack($LONG_PACK, substr($tag->{content}, $index * $LONG_SIZE, $LONG_SIZE) );
370 if (!$location) { return; }
372 return $self->_load_tag( $location );
377 # Adds one key/value pair to bucket list, given offset, MD5 digest of key,
378 # plain (undigested) key and value.
381 my ($tag, $md5, $plain_key, $value) = @_;
382 my $keys = $tag->{content};
386 # added ref() check first to avoid eval and runtime exception for every
387 # scalar value being stored. performance tweak.
388 my $is_dbm_deep = ref($value) && eval { $value->isa( 'DBM::Deep' ) };
390 my $internal_ref = $is_dbm_deep && ($value->root eq $self->root);
395 # Iterate through buckets, seeing if this is a new entry or a replace.
397 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
398 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
399 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
402 # Found empty bucket (end of list). Populate and exit loop.
406 $location = $internal_ref
407 ? $value->base_offset
408 : $self->root->{end};
410 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
411 print($fh $md5 . pack($LONG_PACK, $location) );
414 elsif ($md5 eq $key) {
416 # Found existing bucket with same key. Replace with new value.
421 $location = $value->base_offset;
422 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
423 print($fh $md5 . pack($LONG_PACK, $location) );
426 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
428 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
431 # If value is a hash, array, or raw value with equal or less size, we can
432 # reuse the same content area of the database. Otherwise, we have to create
433 # a new content area at the EOF.
436 my $r = Scalar::Util::reftype( $value ) || '';
437 if ( $r eq 'HASH' || $r eq 'ARRAY' ) {
438 $actual_length = $INDEX_SIZE;
440 # if autobless is enabled, must also take into consideration
441 # the class name, as it is stored along with key/value.
442 if ( $self->root->{autobless} ) {
443 my $value_class = Scalar::Util::blessed($value);
444 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
445 $actual_length += length($value_class);
449 else { $actual_length = length($value); }
451 if ($actual_length <= $size) {
455 $location = $self->root->{end};
456 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE, SEEK_SET);
457 print($fh pack($LONG_PACK, $location) );
465 # If this is an internal reference, return now.
466 # No need to write value or plain key
473 # If bucket didn't fit into list, split into a new index level
476 seek($fh, $tag->{ref_loc}, SEEK_SET);
477 print($fh pack($LONG_PACK, $self->root->{end}) );
479 my $index_tag = $self->_create_tag($self->root->{end}, SIG_INDEX, chr(0) x $INDEX_SIZE);
482 $keys .= $md5 . pack($LONG_PACK, 0);
484 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
485 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
487 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
488 my $num = ord(substr($key, $tag->{ch} + 1, 1));
490 if ($offsets[$num]) {
491 my $offset = $offsets[$num] + SIG_SIZE + $DATA_LENGTH_SIZE;
492 seek($fh, $offset, SEEK_SET);
494 read( $fh, $subkeys, $BUCKET_LIST_SIZE);
496 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
497 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
499 seek($fh, $offset + ($k * $BUCKET_SIZE), SEEK_SET);
500 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
506 $offsets[$num] = $self->root->{end};
507 seek($fh, $index_tag->{offset} + ($num * $LONG_SIZE), SEEK_SET);
508 print($fh pack($LONG_PACK, $self->root->{end}) );
510 my $blist_tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
512 seek($fh, $blist_tag->{offset}, SEEK_SET);
513 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
518 $location ||= $self->root->{end};
519 } # re-index bucket list
522 # Seek to content area and store signature, value and plaintext key
526 seek($fh, $location, SEEK_SET);
529 # Write signature based on content type, set content length and write actual value.
531 my $r = Scalar::Util::reftype($value) || '';
533 print($fh TYPE_HASH );
534 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
535 $content_length = $INDEX_SIZE;
537 elsif ($r eq 'ARRAY') {
538 print($fh TYPE_ARRAY );
539 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
540 $content_length = $INDEX_SIZE;
542 elsif (!defined($value)) {
543 print($fh SIG_NULL );
544 print($fh pack($DATA_LENGTH_PACK, 0) );
548 print($fh SIG_DATA );
549 print($fh pack($DATA_LENGTH_PACK, length($value)) . $value );
550 $content_length = length($value);
554 # Plain key is stored AFTER value, as keys are typically fetched less often.
556 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
559 # If value is blessed, preserve class name
561 if ( $self->root->{autobless} ) {
562 my $value_class = Scalar::Util::blessed($value);
563 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
565 # Blessed ref -- will restore later
568 print($fh pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
569 $content_length += 1;
570 $content_length += $DATA_LENGTH_SIZE + length($value_class);
574 $content_length += 1;
579 # If this is a new content area, advance EOF counter
581 if ($location == $self->root->{end}) {
582 $self->root->{end} += SIG_SIZE;
583 $self->root->{end} += $DATA_LENGTH_SIZE + $content_length;
584 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
588 # If content is a hash or array, create new child DBM::Deep object and
589 # pass each key or element to it.
592 my $branch = DBM::Deep->new(
594 base_offset => $location,
597 foreach my $key (keys %{$value}) {
598 $branch->STORE( $key, $value->{$key} );
601 elsif ($r eq 'ARRAY') {
602 my $branch = DBM::Deep->new(
604 base_offset => $location,
608 foreach my $element (@{$value}) {
609 $branch->STORE( $index, $element );
617 return $self->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
620 sub _get_bucket_value {
622 # Fetch single value given tag and MD5 digested key.
625 my ($tag, $md5) = @_;
626 my $keys = $tag->{content};
631 # Iterate through buckets, looking for a key match
634 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
635 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
636 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
640 # Hit end of list, no match
645 if ( $md5 ne $key ) {
650 # Found match -- seek to offset and read signature
653 seek($fh, $subloc, SEEK_SET);
654 read( $fh, $signature, SIG_SIZE);
657 # If value is a hash or array, return new DBM::Deep object with correct offset
659 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
660 my $obj = DBM::Deep->new(
662 base_offset => $subloc,
666 if ($self->root->{autobless}) {
668 # Skip over value and plain key to see if object needs
671 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
674 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
675 if ($size) { seek($fh, $size, SEEK_CUR); }
678 read( $fh, $bless_bit, 1);
679 if (ord($bless_bit)) {
681 # Yes, object needs to be re-blessed
684 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
685 if ($size) { read( $fh, $class_name, $size); }
686 if ($class_name) { $obj = bless( $obj, $class_name ); }
694 # Otherwise return actual value
696 elsif ($signature eq SIG_DATA) {
699 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
700 if ($size) { read( $fh, $value, $size); }
705 # Key exists, but content is null
715 # Delete single key/value pair given tag and MD5 digested key.
718 my ($tag, $md5) = @_;
719 my $keys = $tag->{content};
724 # Iterate through buckets, looking for a key match
727 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
728 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
729 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
733 # Hit end of list, no match
738 if ( $md5 ne $key ) {
743 # Matched key -- delete bucket and return
745 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
746 print($fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
747 print($fh chr(0) x $BUCKET_SIZE );
757 # Check existence of single key given tag and MD5 digested key.
760 my ($tag, $md5) = @_;
761 my $keys = $tag->{content};
764 # Iterate through buckets, looking for a key match
767 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
768 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
769 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
773 # Hit end of list, no match
778 if ( $md5 ne $key ) {
783 # Matched key -- return true
791 sub _find_bucket_list {
793 # Locate offset for bucket list, given digested key
799 # Locate offset for bucket list using digest index system
802 my $tag = $self->_load_tag($self->base_offset);
803 if (!$tag) { return; }
805 while ($tag->{signature} ne SIG_BLIST) {
806 $tag = $self->_index_lookup($tag, ord(substr($md5, $ch, 1)));
807 if (!$tag) { return; }
814 sub _traverse_index {
816 # Scan index and recursively step into deeper levels, looking for next key.
818 my ($self, $offset, $ch, $force_return_next) = @_;
819 $force_return_next = undef unless $force_return_next;
821 my $tag = $self->_load_tag( $offset );
825 if ($tag->{signature} ne SIG_BLIST) {
826 my $content = $tag->{content};
828 if ($self->{return_next}) { $start = 0; }
829 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
831 for (my $index = $start; $index < 256; $index++) {
832 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
834 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
835 if (defined($result)) { return $result; }
839 $self->{return_next} = 1;
842 elsif ($tag->{signature} eq SIG_BLIST) {
843 my $keys = $tag->{content};
844 if ($force_return_next) { $self->{return_next} = 1; }
847 # Iterate through buckets, looking for a key match
849 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
850 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
851 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
855 # End of bucket list -- return to outer loop
857 $self->{return_next} = 1;
860 elsif ($key eq $self->{prev_md5}) {
862 # Located previous key -- return next one found
864 $self->{return_next} = 1;
867 elsif ($self->{return_next}) {
869 # Seek to bucket location and skip over signature
871 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
874 # Skip over value to get to plain key
877 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
878 if ($size) { seek($fh, $size, SEEK_CUR); }
881 # Read in plain key and return as scalar
884 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
885 if ($size) { read( $fh, $plain_key, $size); }
891 $self->{return_next} = 1;
892 } # tag is a bucket list
899 # Locate next key, given digested previous one
901 my $self = $_[0]->_get_self;
903 $self->{prev_md5} = $_[1] ? $_[1] : undef;
904 $self->{return_next} = 0;
907 # If the previous key was not specifed, start at the top and
908 # return the first one found.
910 if (!$self->{prev_md5}) {
911 $self->{prev_md5} = chr(0) x $HASH_SIZE;
912 $self->{return_next} = 1;
915 return $self->_traverse_index( $self->base_offset, 0 );
920 # If db locking is set, flock() the db file. If called multiple
921 # times before unlock(), then the same number of unlocks() must
922 # be called before the lock is released.
924 my $self = $_[0]->_get_self;
926 $type = LOCK_EX unless defined $type;
928 if (!defined($self->fh)) { return; }
930 if ($self->root->{locking}) {
931 if (!$self->root->{locked}) {
932 flock($self->fh, $type);
934 # refresh end counter in case file has changed size
935 my @stats = stat($self->root->{file});
936 $self->root->{end} = $stats[7];
938 # double-check file inode, in case another process
939 # has optimize()d our file while we were waiting.
940 if ($stats[1] != $self->root->{inode}) {
941 $self->_open(); # re-open
942 flock($self->fh, $type); # re-lock
943 $self->root->{end} = (stat($self->fh))[7]; # re-end
946 $self->root->{locked}++;
956 # If db locking is set, unlock the db file. See note in lock()
957 # regarding calling lock() multiple times.
959 my $self = $_[0]->_get_self;
961 if (!defined($self->fh)) { return; }
963 if ($self->root->{locking} && $self->root->{locked} > 0) {
964 $self->root->{locked}--;
965 if (!$self->root->{locked}) { flock($self->fh, LOCK_UN); }
973 #XXX These uses of ref() need verified
976 # Copy single level of keys or elements to new DB handle.
977 # Recurse for nested structures
979 my $self = $_[0]->_get_self;
982 if ($self->type eq TYPE_HASH) {
983 my $key = $self->first_key();
985 my $value = $self->get($key);
986 #XXX This doesn't work with autobless
987 if (!ref($value)) { $db_temp->{$key} = $value; }
989 my $type = $value->type;
990 if ($type eq TYPE_HASH) { $db_temp->{$key} = {}; }
991 else { $db_temp->{$key} = []; }
992 $value->_copy_node( $db_temp->{$key} );
994 $key = $self->next_key($key);
998 my $length = $self->length();
999 for (my $index = 0; $index < $length; $index++) {
1000 my $value = $self->get($index);
1001 if (!ref($value)) { $db_temp->[$index] = $value; }
1002 #XXX NO tests for this code
1004 my $type = $value->type;
1005 if ($type eq TYPE_HASH) { $db_temp->[$index] = {}; }
1006 else { $db_temp->[$index] = []; }
1007 $value->_copy_node( $db_temp->[$index] );
1015 # Recursively export into standard Perl hashes and arrays.
1017 my $self = $_[0]->_get_self;
1020 if ($self->type eq TYPE_HASH) { $temp = {}; }
1021 elsif ($self->type eq TYPE_ARRAY) { $temp = []; }
1024 $self->_copy_node( $temp );
1032 # Recursively import Perl hash/array structure
1034 #XXX This use of ref() seems to be ok
1035 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
1037 my $self = $_[0]->_get_self;
1040 #XXX This use of ref() seems to be ok
1041 if (!ref($struct)) {
1043 # struct is not a reference, so just import based on our type
1047 if ($self->type eq TYPE_HASH) { $struct = {@_}; }
1048 elsif ($self->type eq TYPE_ARRAY) { $struct = [@_]; }
1051 my $r = Scalar::Util::reftype($struct) || '';
1052 if ($r eq "HASH" && $self->type eq TYPE_HASH) {
1053 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
1055 elsif ($r eq "ARRAY" && $self->type eq TYPE_ARRAY) {
1056 $self->push( @$struct );
1059 return $self->_throw_error("Cannot import: type mismatch");
1067 # Rebuild entire database into new file, then move
1068 # it back on top of original.
1070 my $self = $_[0]->_get_self;
1072 #XXX Need to create a new test for this
1073 # if ($self->root->{links} > 1) {
1074 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
1077 my $db_temp = DBM::Deep->new(
1078 file => $self->root->{file} . '.tmp',
1082 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
1086 $self->_copy_node( $db_temp );
1090 # Attempt to copy user, group and permissions over to new file
1092 my @stats = stat($self->fh);
1093 my $perms = $stats[2] & 07777;
1094 my $uid = $stats[4];
1095 my $gid = $stats[5];
1096 chown( $uid, $gid, $self->root->{file} . '.tmp' );
1097 chmod( $perms, $self->root->{file} . '.tmp' );
1099 # q.v. perlport for more information on this variable
1100 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
1102 # Potential race condition when optmizing on Win32 with locking.
1103 # The Windows filesystem requires that the filehandle be closed
1104 # before it is overwritten with rename(). This could be redone
1111 if (!rename $self->root->{file} . '.tmp', $self->root->{file}) {
1112 unlink $self->root->{file} . '.tmp';
1114 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
1126 # Make copy of object and return
1128 my $self = $_[0]->_get_self;
1130 return DBM::Deep->new(
1131 type => $self->type,
1132 base_offset => $self->base_offset,
1138 my %is_legal_filter = map {
1141 store_key store_value
1142 fetch_key fetch_value
1147 # Setup filter function for storing or fetching the key or value
1149 my $self = $_[0]->_get_self;
1150 my $type = lc $_[1];
1151 my $func = $_[2] ? $_[2] : undef;
1153 if ( $is_legal_filter{$type} ) {
1154 $self->root->{"filter_$type"} = $func;
1168 # Get access to the root structure
1170 my $self = $_[0]->_get_self;
1171 return $self->{root};
1176 # Get access to the raw fh
1178 #XXX It will be useful, though, when we split out HASH and ARRAY
1179 my $self = $_[0]->_get_self;
1180 return $self->root->{fh};
1185 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
1187 my $self = $_[0]->_get_self;
1188 return $self->{type};
1193 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
1195 my $self = $_[0]->_get_self;
1196 return $self->{base_offset};
1201 # Get last error string, or undef if no error
1204 #? ( _get_self($_[0])->{root}->{error} or undef )
1205 ? ( $_[0]->_get_self->{root}->{error} or undef )
1215 # Store error string in self
1217 my $self = $_[0]->_get_self;
1218 my $error_text = $_[1];
1220 if ( Scalar::Util::blessed $self ) {
1221 $self->root->{error} = $error_text;
1223 unless ($self->root->{debug}) {
1224 die "DBM::Deep: $error_text\n";
1227 warn "DBM::Deep: $error_text\n";
1231 die "DBM::Deep: $error_text\n";
1239 my $self = $_[0]->_get_self;
1241 undef $self->root->{error};
1246 # Precalculate index, bucket and bucket list sizes
1249 #XXX I don't like this ...
1250 set_pack() unless defined $LONG_SIZE;
1252 $INDEX_SIZE = 256 * $LONG_SIZE;
1253 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
1254 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
1259 # Set pack/unpack modes (see file header for more)
1261 my ($long_s, $long_p, $data_s, $data_p) = @_;
1263 $LONG_SIZE = $long_s ? $long_s : 4;
1264 $LONG_PACK = $long_p ? $long_p : 'N';
1266 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
1267 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
1274 # Set key digest function (default is MD5)
1276 my ($digest_func, $hash_size) = @_;
1278 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
1279 $HASH_SIZE = $hash_size ? $hash_size : 16;
1285 # tie() methods (hashes and arrays)
1290 # Store single hash key/value or array element in database.
1292 my $self = $_[0]->_get_self;
1295 # User may be storing a hash, in which case we do not want it run
1296 # through the filtering system
1297 my $value = ($self->root->{filter_store_value} && !ref($_[2]))
1298 ? $self->root->{filter_store_value}->($_[2])
1301 my $md5 = $DIGEST_FUNC->($key);
1304 # Make sure file is open
1306 if (!defined($self->fh) && !$self->_open()) {
1312 # Request exclusive lock for writing
1314 $self->lock( LOCK_EX );
1319 # Locate offset for bucket list using digest index system
1321 my $tag = $self->_load_tag($self->base_offset);
1323 $tag = $self->_create_tag($self->base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1327 while ($tag->{signature} ne SIG_BLIST) {
1328 my $num = ord(substr($md5, $ch, 1));
1329 my $new_tag = $self->_index_lookup($tag, $num);
1331 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1332 seek($fh, $ref_loc, SEEK_SET);
1333 print($fh pack($LONG_PACK, $self->root->{end}) );
1335 $tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1336 $tag->{ref_loc} = $ref_loc;
1341 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1343 $tag->{ref_loc} = $ref_loc;
1350 # Add key/value to bucket list
1352 my $result = $self->_add_bucket( $tag, $md5, $key, $value );
1361 # Fetch single value or element given plain key or array index
1363 my $self = shift->_get_self;
1367 # Make sure file is open
1369 if (!defined($self->fh)) { $self->_open(); }
1371 my $md5 = $DIGEST_FUNC->($key);
1374 # Request shared lock for reading
1376 $self->lock( LOCK_SH );
1378 my $tag = $self->_find_bucket_list( $md5 );
1385 # Get value from bucket list
1387 my $result = $self->_get_bucket_value( $tag, $md5 );
1391 #XXX What is ref() checking here?
1392 #YYY Filters only apply on scalar values, so the ref check is making
1393 #YYY sure the fetched bucket is a scalar, not a child hash or array.
1394 return ($result && !ref($result) && $self->root->{filter_fetch_value})
1395 ? $self->root->{filter_fetch_value}->($result)
1401 # Delete single key/value pair or element given plain key or array index
1403 my $self = $_[0]->_get_self;
1406 my $md5 = $DIGEST_FUNC->($key);
1409 # Make sure file is open
1411 if (!defined($self->fh)) { $self->_open(); }
1414 # Request exclusive lock for writing
1416 $self->lock( LOCK_EX );
1418 my $tag = $self->_find_bucket_list( $md5 );
1427 my $value = $self->_get_bucket_value( $tag, $md5 );
1428 if ($value && !ref($value) && $self->root->{filter_fetch_value}) {
1429 $value = $self->root->{filter_fetch_value}->($value);
1432 my $result = $self->_delete_bucket( $tag, $md5 );
1435 # If this object is an array and the key deleted was on the end of the stack,
1436 # decrement the length variable.
1446 # Check if a single key or element exists given plain key or array index
1448 my $self = $_[0]->_get_self;
1451 my $md5 = $DIGEST_FUNC->($key);
1454 # Make sure file is open
1456 if (!defined($self->fh)) { $self->_open(); }
1459 # Request shared lock for reading
1461 $self->lock( LOCK_SH );
1463 my $tag = $self->_find_bucket_list( $md5 );
1466 # For some reason, the built-in exists() function returns '' for false
1474 # Check if bucket exists and return 1 or ''
1476 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1485 # Clear all keys from hash, or all elements from array.
1487 my $self = $_[0]->_get_self;
1490 # Make sure file is open
1492 if (!defined($self->fh)) { $self->_open(); }
1495 # Request exclusive lock for writing
1497 $self->lock( LOCK_EX );
1501 seek($fh, $self->base_offset, SEEK_SET);
1507 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
1515 # Public method aliases
1517 sub put { (shift)->STORE( @_ ) }
1518 sub store { (shift)->STORE( @_ ) }
1519 sub get { (shift)->FETCH( @_ ) }
1520 sub fetch { (shift)->FETCH( @_ ) }
1521 sub delete { (shift)->DELETE( @_ ) }
1522 sub exists { (shift)->EXISTS( @_ ) }
1523 sub clear { (shift)->CLEAR( @_ ) }
1525 package DBM::Deep::_::Root;
1538 filter_store_key => undef,
1539 filter_store_value => undef,
1540 filter_fetch_key => undef,
1541 filter_fetch_value => undef,
1552 return unless $self;
1554 close $self->{fh} if $self->{fh};
1565 DBM::Deep - A pure perl multi-level hash/array DBM
1570 my $db = DBM::Deep->new( "foo.db" );
1572 $db->{key} = 'value'; # tie() style
1575 $db->put('key' => 'value'); # OO style
1576 print $db->get('key');
1578 # true multi-level support
1579 $db->{my_complex} = [
1580 'hello', { perl => 'rules' },
1586 A unique flat-file database module, written in pure perl. True
1587 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1588 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1589 handle millions of keys and unlimited hash levels without significant
1590 slow-down. Written from the ground-up in pure perl -- this is NOT a
1591 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1592 Mac OS X and Windows.
1596 Hopefully you are using Perl's excellent CPAN module, which will download
1597 and install the module for you. If not, get the tarball, and run these
1609 Construction can be done OO-style (which is the recommended way), or using
1610 Perl's tie() function. Both are examined here.
1612 =head2 OO CONSTRUCTION
1614 The recommended way to construct a DBM::Deep object is to use the new()
1615 method, which gets you a blessed, tied hash or array reference.
1617 my $db = DBM::Deep->new( "foo.db" );
1619 This opens a new database handle, mapped to the file "foo.db". If this
1620 file does not exist, it will automatically be created. DB files are
1621 opened in "r+" (read/write) mode, and the type of object returned is a
1622 hash, unless otherwise specified (see L<OPTIONS> below).
1624 You can pass a number of options to the constructor to specify things like
1625 locking, autoflush, etc. This is done by passing an inline hash:
1627 my $db = DBM::Deep->new(
1633 Notice that the filename is now specified I<inside> the hash with
1634 the "file" parameter, as opposed to being the sole argument to the
1635 constructor. This is required if any options are specified.
1636 See L<OPTIONS> below for the complete list.
1640 You can also start with an array instead of a hash. For this, you must
1641 specify the C<type> parameter:
1643 my $db = DBM::Deep->new(
1645 type => DBM::Deep->TYPE_ARRAY
1648 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1649 a new DB file. If you create a DBM::Deep object with an existing file, the
1650 C<type> will be loaded from the file header, and an error will be thrown if
1651 the wrong type is passed in.
1653 =head2 TIE CONSTRUCTION
1655 Alternately, you can create a DBM::Deep handle by using Perl's built-in
1656 tie() function. The object returned from tie() can be used to call methods,
1657 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
1658 file (as expected with most tie'd objects).
1661 my $db = tie %hash, "DBM::Deep", "foo.db";
1664 my $db = tie @array, "DBM::Deep", "bar.db";
1666 As with the OO constructor, you can replace the DB filename parameter with
1667 a hash containing one or more options (see L<OPTIONS> just below for the
1670 tie %hash, "DBM::Deep", {
1678 There are a number of options that can be passed in when constructing your
1679 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1685 Filename of the DB file to link the handle to. You can pass a full absolute
1686 filesystem path, partial path, or a plain filename if the file is in the
1687 current working directory. This is a required parameter.
1691 This parameter specifies what type of object to create, a hash or array. Use
1692 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1693 This only takes effect when beginning a new file. This is an optional
1694 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1698 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1699 function to lock the database in exclusive mode for writes, and shared mode for
1700 reads. Pass any true value to enable. This affects the base DB handle I<and
1701 any child hashes or arrays> that use the same DB file. This is an optional
1702 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1706 Specifies whether autoflush is to be enabled on the underlying filehandle.
1707 This obviously slows down write operations, but is required if you may have
1708 multiple processes accessing the same DB file (also consider enable I<locking>).
1709 Pass any true value to enable. This is an optional parameter, and defaults to 0
1714 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1715 restore them when fetched. This is an B<experimental> feature, and does have
1716 side-effects. Basically, when hashes are re-blessed into their original
1717 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1718 able to call any DBM::Deep methods on them. You have been warned.
1719 This is an optional parameter, and defaults to 0 (disabled).
1723 See L<FILTERS> below.
1727 Setting I<debug> mode will make all errors non-fatal, dump them out to
1728 STDERR, and continue on. This is for debugging purposes only, and probably
1729 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1733 Instead of passing a file path, you can instead pass a handle to an pre-opened
1734 filehandle. Note: Beware of using the magick *DATA handle, as this actually
1735 contains your entire Perl script, as well as the data following the __DATA__
1736 marker. This will not work, because DBM::Deep uses absolute seek()s into the
1737 file. Instead, consider reading *DATA into an IO::Scalar handle, then passing
1738 in that. Also please note optimize() will NOT work when passing in only a
1739 handle. Pass in a real filename in order to use optimize().
1743 =head1 TIE INTERFACE
1745 With DBM::Deep you can access your databases using Perl's standard hash/array
1746 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
1747 treat them as such. DBM::Deep will intercept all reads/writes and direct them
1748 to the right place -- the DB file. This has nothing to do with the
1749 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
1750 using regular hashes and arrays, rather than calling functions like C<get()>
1751 and C<put()> (although those work too). It is entirely up to you how to want
1752 to access your databases.
1756 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1757 or even nested hashes (or arrays) using standard Perl syntax:
1759 my $db = DBM::Deep->new( "foo.db" );
1761 $db->{mykey} = "myvalue";
1763 $db->{myhash}->{subkey} = "subvalue";
1765 print $db->{myhash}->{subkey} . "\n";
1767 You can even step through hash keys using the normal Perl C<keys()> function:
1769 foreach my $key (keys %$db) {
1770 print "$key: " . $db->{$key} . "\n";
1773 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1774 pushes them onto an array, all before the loop even begins. If you have an
1775 extra large hash, this may exhaust Perl's memory. Instead, consider using
1776 Perl's C<each()> function, which pulls keys/values one at a time, using very
1779 while (my ($key, $value) = each %$db) {
1780 print "$key: $value\n";
1783 Please note that when using C<each()>, you should always pass a direct
1784 hash reference, not a lookup. Meaning, you should B<never> do this:
1787 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1789 This causes an infinite loop, because for each iteration, Perl is calling
1790 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1791 it effectively keeps returning the first key over and over again. Instead,
1792 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1796 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1797 reference. This includes inserting, removing and manipulating elements,
1798 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1799 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1800 or simply be a nested array reference inside a hash. Example:
1802 my $db = DBM::Deep->new(
1803 file => "foo-array.db",
1804 type => DBM::Deep->TYPE_ARRAY
1808 push @$db, "bar", "baz";
1809 unshift @$db, "bah";
1811 my $last_elem = pop @$db; # baz
1812 my $first_elem = shift @$db; # bah
1813 my $second_elem = $db->[1]; # bar
1815 my $num_elements = scalar @$db;
1819 In addition to the I<tie()> interface, you can also use a standard OO interface
1820 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1821 array) has its own methods, but both types share the following common methods:
1822 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1826 =item * put() / store()
1828 Stores a new hash key/value pair, or sets an array element value. Takes two
1829 arguments, the hash key or array index, and the new value. The value can be
1830 a scalar, hash ref or array ref. Returns true on success, false on failure.
1832 $db->put("foo", "bar"); # for hashes
1833 $db->put(1, "bar"); # for arrays
1835 =item * get() / fetch()
1837 Fetches the value of a hash key or array element. Takes one argument: the hash
1838 key or array index. Returns a scalar, hash ref or array ref, depending on the
1841 my $value = $db->get("foo"); # for hashes
1842 my $value = $db->get(1); # for arrays
1846 Checks if a hash key or array index exists. Takes one argument: the hash key
1847 or array index. Returns true if it exists, false if not.
1849 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1850 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1854 Deletes one hash key/value pair or array element. Takes one argument: the hash
1855 key or array index. Returns true on success, false if not found. For arrays,
1856 the remaining elements located after the deleted element are NOT moved over.
1857 The deleted element is essentially just undefined, which is exactly how Perl's
1858 internal arrays work. Please note that the space occupied by the deleted
1859 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1860 below for details and workarounds.
1862 $db->delete("foo"); # for hashes
1863 $db->delete(1); # for arrays
1867 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1868 value. Please note that the space occupied by the deleted keys/values or
1869 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1870 details and workarounds.
1872 $db->clear(); # hashes or arrays
1878 For hashes, DBM::Deep supports all the common methods described above, and the
1879 following additional methods: C<first_key()> and C<next_key()>.
1885 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1886 fetched in an undefined order (which appears random). Takes no arguments,
1887 returns the key as a scalar value.
1889 my $key = $db->first_key();
1893 Returns the "next" key in the hash, given the previous one as the sole argument.
1894 Returns undef if there are no more keys to be fetched.
1896 $key = $db->next_key($key);
1900 Here are some examples of using hashes:
1902 my $db = DBM::Deep->new( "foo.db" );
1904 $db->put("foo", "bar");
1905 print "foo: " . $db->get("foo") . "\n";
1907 $db->put("baz", {}); # new child hash ref
1908 $db->get("baz")->put("buz", "biz");
1909 print "buz: " . $db->get("baz")->get("buz") . "\n";
1911 my $key = $db->first_key();
1913 print "$key: " . $db->get($key) . "\n";
1914 $key = $db->next_key($key);
1917 if ($db->exists("foo")) { $db->delete("foo"); }
1921 For arrays, DBM::Deep supports all the common methods described above, and the
1922 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1923 C<unshift()> and C<splice()>.
1929 Returns the number of elements in the array. Takes no arguments.
1931 my $len = $db->length();
1935 Adds one or more elements onto the end of the array. Accepts scalars, hash
1936 refs or array refs. No return value.
1938 $db->push("foo", "bar", {});
1942 Fetches the last element in the array, and deletes it. Takes no arguments.
1943 Returns undef if array is empty. Returns the element value.
1945 my $elem = $db->pop();
1949 Fetches the first element in the array, deletes it, then shifts all the
1950 remaining elements over to take up the space. Returns the element value. This
1951 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1954 my $elem = $db->shift();
1958 Inserts one or more elements onto the beginning of the array, shifting all
1959 existing elements over to make room. Accepts scalars, hash refs or array refs.
1960 No return value. This method is not recommended with large arrays -- see
1961 <LARGE ARRAYS> below for details.
1963 $db->unshift("foo", "bar", {});
1967 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1968 -f splice> for usage -- it is too complicated to document here. This method is
1969 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1973 Here are some examples of using arrays:
1975 my $db = DBM::Deep->new(
1977 type => DBM::Deep->TYPE_ARRAY
1980 $db->push("bar", "baz");
1981 $db->unshift("foo");
1984 my $len = $db->length();
1985 print "length: $len\n"; # 4
1987 for (my $k=0; $k<$len; $k++) {
1988 print "$k: " . $db->get($k) . "\n";
1991 $db->splice(1, 2, "biz", "baf");
1993 while (my $elem = shift @$db) {
1994 print "shifted: $elem\n";
1999 Enable automatic file locking by passing a true value to the C<locking>
2000 parameter when constructing your DBM::Deep object (see L<SETUP> above).
2002 my $db = DBM::Deep->new(
2007 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
2008 mode for writes, and shared mode for reads. This is required if you have
2009 multiple processes accessing the same database file, to avoid file corruption.
2010 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
2011 NFS> below for more.
2013 =head2 EXPLICIT LOCKING
2015 You can explicitly lock a database, so it remains locked for multiple
2016 transactions. This is done by calling the C<lock()> method, and passing an
2017 optional lock mode argument (defaults to exclusive mode). This is particularly
2018 useful for things like counters, where the current value needs to be fetched,
2019 then incremented, then stored again.
2022 my $counter = $db->get("counter");
2024 $db->put("counter", $counter);
2033 You can pass C<lock()> an optional argument, which specifies which mode to use
2034 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2035 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2036 same as the constants defined in Perl's C<Fcntl> module.
2038 $db->lock( DBM::Deep->LOCK_SH );
2042 =head1 IMPORTING/EXPORTING
2044 You can import existing complex structures by calling the C<import()> method,
2045 and export an entire database into an in-memory structure using the C<export()>
2046 method. Both are examined here.
2050 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2051 walking the structure and adding keys/elements to the database as you go,
2052 simply pass a reference to the C<import()> method. This recursively adds
2053 everything to an existing DBM::Deep object for you. Here is an example:
2058 array1 => [ "elem0", "elem1", "elem2" ],
2060 subkey1 => "subvalue1",
2061 subkey2 => "subvalue2"
2065 my $db = DBM::Deep->new( "foo.db" );
2066 $db->import( $struct );
2068 print $db->{key1} . "\n"; # prints "value1"
2070 This recursively imports the entire C<$struct> object into C<$db>, including
2071 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2072 keys are merged with the existing ones, replacing if they already exist.
2073 The C<import()> method can be called on any database level (not just the base
2074 level), and works with both hash and array DB types.
2076 B<Note:> Make sure your existing structure has no circular references in it.
2077 These will cause an infinite loop when importing.
2081 Calling the C<export()> method on an existing DBM::Deep object will return
2082 a reference to a new in-memory copy of the database. The export is done
2083 recursively, so all nested hashes/arrays are all exported to standard Perl
2084 objects. Here is an example:
2086 my $db = DBM::Deep->new( "foo.db" );
2088 $db->{key1} = "value1";
2089 $db->{key2} = "value2";
2091 $db->{hash1}->{subkey1} = "subvalue1";
2092 $db->{hash1}->{subkey2} = "subvalue2";
2094 my $struct = $db->export();
2096 print $struct->{key1} . "\n"; # prints "value1"
2098 This makes a complete copy of the database in memory, and returns a reference
2099 to it. The C<export()> method can be called on any database level (not just
2100 the base level), and works with both hash and array DB types. Be careful of
2101 large databases -- you can store a lot more data in a DBM::Deep object than an
2102 in-memory Perl structure.
2104 B<Note:> Make sure your database has no circular references in it.
2105 These will cause an infinite loop when exporting.
2109 DBM::Deep has a number of hooks where you can specify your own Perl function
2110 to perform filtering on incoming or outgoing data. This is a perfect
2111 way to extend the engine, and implement things like real-time compression or
2112 encryption. Filtering applies to the base DB level, and all child hashes /
2113 arrays. Filter hooks can be specified when your DBM::Deep object is first
2114 constructed, or by calling the C<set_filter()> method at any time. There are
2115 four available filter hooks, described below:
2119 =item * filter_store_key
2121 This filter is called whenever a hash key is stored. It
2122 is passed the incoming key, and expected to return a transformed key.
2124 =item * filter_store_value
2126 This filter is called whenever a hash key or array element is stored. It
2127 is passed the incoming value, and expected to return a transformed value.
2129 =item * filter_fetch_key
2131 This filter is called whenever a hash key is fetched (i.e. via
2132 C<first_key()> or C<next_key()>). It is passed the transformed key,
2133 and expected to return the plain key.
2135 =item * filter_fetch_value
2137 This filter is called whenever a hash key or array element is fetched.
2138 It is passed the transformed value, and expected to return the plain value.
2142 Here are the two ways to setup a filter hook:
2144 my $db = DBM::Deep->new(
2146 filter_store_value => \&my_filter_store,
2147 filter_fetch_value => \&my_filter_fetch
2152 $db->set_filter( "filter_store_value", \&my_filter_store );
2153 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2155 Your filter function will be called only when dealing with SCALAR keys or
2156 values. When nested hashes and arrays are being stored/fetched, filtering
2157 is bypassed. Filters are called as static functions, passed a single SCALAR
2158 argument, and expected to return a single SCALAR value. If you want to
2159 remove a filter, set the function reference to C<undef>:
2161 $db->set_filter( "filter_store_value", undef );
2163 =head2 REAL-TIME ENCRYPTION EXAMPLE
2165 Here is a working example that uses the I<Crypt::Blowfish> module to
2166 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2167 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2168 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2171 use Crypt::Blowfish;
2174 my $cipher = Crypt::CBC->new({
2175 'key' => 'my secret key',
2176 'cipher' => 'Blowfish',
2178 'regenerate_key' => 0,
2179 'padding' => 'space',
2183 my $db = DBM::Deep->new(
2184 file => "foo-encrypt.db",
2185 filter_store_key => \&my_encrypt,
2186 filter_store_value => \&my_encrypt,
2187 filter_fetch_key => \&my_decrypt,
2188 filter_fetch_value => \&my_decrypt,
2191 $db->{key1} = "value1";
2192 $db->{key2} = "value2";
2193 print "key1: " . $db->{key1} . "\n";
2194 print "key2: " . $db->{key2} . "\n";
2200 return $cipher->encrypt( $_[0] );
2203 return $cipher->decrypt( $_[0] );
2206 =head2 REAL-TIME COMPRESSION EXAMPLE
2208 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2209 compression / decompression of keys & values with DBM::Deep Filters.
2210 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2211 more on I<Compress::Zlib>.
2216 my $db = DBM::Deep->new(
2217 file => "foo-compress.db",
2218 filter_store_key => \&my_compress,
2219 filter_store_value => \&my_compress,
2220 filter_fetch_key => \&my_decompress,
2221 filter_fetch_value => \&my_decompress,
2224 $db->{key1} = "value1";
2225 $db->{key2} = "value2";
2226 print "key1: " . $db->{key1} . "\n";
2227 print "key2: " . $db->{key2} . "\n";
2233 return Compress::Zlib::memGzip( $_[0] ) ;
2236 return Compress::Zlib::memGunzip( $_[0] ) ;
2239 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2240 actually numerical index numbers, and are not filtered.
2242 =head1 ERROR HANDLING
2244 Most DBM::Deep methods return a true value for success, and call die() on
2245 failure. You can wrap calls in an eval block to catch the die. Also, the
2246 actual error message is stored in an internal scalar, which can be fetched by
2247 calling the C<error()> method.
2249 my $db = DBM::Deep->new( "foo.db" ); # create hash
2250 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2252 print $@; # prints error message
2253 print $db->error(); # prints error message
2255 You can then call C<clear_error()> to clear the current error state.
2259 If you set the C<debug> option to true when creating your DBM::Deep object,
2260 all errors are considered NON-FATAL, and dumped to STDERR. This should only
2261 be used for debugging purposes and not production work. DBM::Deep expects errors
2262 to be thrown, not propagated back up the stack.
2264 =head1 LARGEFILE SUPPORT
2266 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2267 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2268 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2269 by calling the static C<set_pack()> method before you do anything else.
2271 DBM::Deep::set_pack(8, 'Q');
2273 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2274 instead of 32-bit longs. After setting these values your DB files have a
2275 theoretical maximum size of 16 XB (exabytes).
2277 B<Note:> Changing these values will B<NOT> work for existing database files.
2278 Only change this for new files, and make sure it stays set consistently
2279 throughout the file's life. If you do set these values, you can no longer
2280 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2281 back to 32-bit mode.
2283 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2284 only a 32-bit Perl. However, I have received user reports that this does
2287 =head1 LOW-LEVEL ACCESS
2289 If you require low-level access to the underlying filehandle that DBM::Deep uses,
2290 you can call the C<fh()> method, which returns the handle:
2294 This method can be called on the root level of the datbase, or any child
2295 hashes or arrays. All levels share a I<root> structure, which contains things
2296 like the filehandle, a reference counter, and all the options specified
2297 when you created the object. You can get access to this root structure by
2298 calling the C<root()> method.
2300 my $root = $db->root();
2302 This is useful for changing options after the object has already been created,
2303 such as enabling/disabling locking, or debug modes. You can also
2304 store your own temporary user data in this structure (be wary of name
2305 collision), which is then accessible from any child hash or array.
2307 =head1 CUSTOM DIGEST ALGORITHM
2309 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2310 keys. However you can override this, and use another algorithm (such as SHA-256)
2311 or even write your own. But please note that DBM::Deep currently expects zero
2312 collisions, so your algorithm has to be I<perfect>, so to speak.
2313 Collision detection may be introduced in a later version.
2317 You can specify a custom digest algorithm by calling the static C<set_digest()>
2318 function, passing a reference to a subroutine, and the length of the algorithm's
2319 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2320 objects. Here is a working example that uses a 256-bit hash from the
2321 I<Digest::SHA256> module. Please see
2322 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2327 my $context = Digest::SHA256::new(256);
2329 DBM::Deep::set_digest( \&my_digest, 32 );
2331 my $db = DBM::Deep->new( "foo-sha.db" );
2333 $db->{key1} = "value1";
2334 $db->{key2} = "value2";
2335 print "key1: " . $db->{key1} . "\n";
2336 print "key2: " . $db->{key2} . "\n";
2342 return substr( $context->hash($_[0]), 0, 32 );
2345 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2346 of bytes you specify in the C<set_digest()> function (in this case 32).
2348 =head1 CIRCULAR REFERENCES
2350 DBM::Deep has B<experimental> support for circular references. Meaning you
2351 can have a nested hash key or array element that points to a parent object.
2352 This relationship is stored in the DB file, and is preserved between sessions.
2355 my $db = DBM::Deep->new( "foo.db" );
2358 $db->{circle} = $db; # ref to self
2360 print $db->{foo} . "\n"; # prints "foo"
2361 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2363 One catch is, passing the object to a function that recursively walks the
2364 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2365 C<export()> methods) will result in an infinite loop. The other catch is,
2366 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2367 or C<next_key()> methods), you will get the I<target object's key>, not the
2368 ref's key. This gets even more interesting with the above example, where
2369 the I<circle> key points to the base DB object, which technically doesn't
2370 have a key. So I made DBM::Deep return "[base]" as the key name in that
2373 =head1 CAVEATS / ISSUES / BUGS
2375 This section describes all the known issues with DBM::Deep. It you have found
2376 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2378 =head2 UNUSED SPACE RECOVERY
2380 One major caveat with DBM::Deep is that space occupied by existing keys and
2381 values is not recovered when they are deleted. Meaning if you keep deleting
2382 and adding new keys, your file will continuously grow. I am working on this,
2383 but in the meantime you can call the built-in C<optimize()> method from time to
2384 time (perhaps in a crontab or something) to recover all your unused space.
2386 $db->optimize(); # returns true on success
2388 This rebuilds the ENTIRE database into a new file, then moves it on top of
2389 the original. The new file will have no unused space, thus it will take up as
2390 little disk space as possible. Please note that this operation can take
2391 a long time for large files, and you need enough disk space to temporarily hold
2392 2 copies of your DB file. The temporary file is created in the same directory
2393 as the original, named with a ".tmp" extension, and is deleted when the
2394 operation completes. Oh, and if locking is enabled, the DB is automatically
2395 locked for the entire duration of the copy.
2397 B<WARNING:> Only call optimize() on the top-level node of the database, and
2398 make sure there are no child references lying around. DBM::Deep keeps a reference
2399 counter, and if it is greater than 1, optimize() will abort and return undef.
2401 =head2 AUTOVIVIFICATION
2403 Unfortunately, autovivification doesn't work with tied hashes. This appears to
2404 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
2405 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
2406 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
2407 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
2410 $db->{foo}->{bar} = "hello";
2412 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
2413 being an empty hash. Try this instead, which works fine:
2415 $db->{foo} = { bar => "hello" };
2417 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
2418 the execution path, and Perl indeed passes an empty hash to the STORE() method.
2419 Probably a bug in Perl.
2421 =head2 FILE CORRUPTION
2423 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2424 for a 32-bit signature when opened, but other corruption in files can cause
2425 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2426 stuck in an infinite loop depending on the level of corruption. File write
2427 operations are not checked for failure (for speed), so if you happen to run
2428 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2429 be addressed in a later version of DBM::Deep.
2433 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2434 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2435 about setting up your NFS server with a locking daemon, then using lockf() to
2436 lock your files, but your mileage may vary there as well. From what I
2437 understand, there is no real way to do it. However, if you need access to the
2438 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
2439 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2441 =head2 COPYING OBJECTS
2443 Beware of copying tied objects in Perl. Very strange things can happen.
2444 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2445 returns a new, blessed, tied hash or array to the same level in the DB.
2447 my $copy = $db->clone();
2449 B<Note>: Since clone() here is cloning the object, not the database location, any
2450 modifications to either $db or $copy will be visible in both.
2454 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2455 These functions cause every element in the array to move, which can be murder
2456 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2457 a different location. This will be addressed in the forthcoming version 1.00.
2461 This section discusses DBM::Deep's speed and memory usage.
2465 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2466 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2467 multi-level hash/array support, and cross-platform FTPable files. Even so,
2468 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2469 with huge databases. Here is some test data:
2471 Adding 1,000,000 keys to new DB file...
2473 At 100 keys, avg. speed is 2,703 keys/sec
2474 At 200 keys, avg. speed is 2,642 keys/sec
2475 At 300 keys, avg. speed is 2,598 keys/sec
2476 At 400 keys, avg. speed is 2,578 keys/sec
2477 At 500 keys, avg. speed is 2,722 keys/sec
2478 At 600 keys, avg. speed is 2,628 keys/sec
2479 At 700 keys, avg. speed is 2,700 keys/sec
2480 At 800 keys, avg. speed is 2,607 keys/sec
2481 At 900 keys, avg. speed is 2,190 keys/sec
2482 At 1,000 keys, avg. speed is 2,570 keys/sec
2483 At 2,000 keys, avg. speed is 2,417 keys/sec
2484 At 3,000 keys, avg. speed is 1,982 keys/sec
2485 At 4,000 keys, avg. speed is 1,568 keys/sec
2486 At 5,000 keys, avg. speed is 1,533 keys/sec
2487 At 6,000 keys, avg. speed is 1,787 keys/sec
2488 At 7,000 keys, avg. speed is 1,977 keys/sec
2489 At 8,000 keys, avg. speed is 2,028 keys/sec
2490 At 9,000 keys, avg. speed is 2,077 keys/sec
2491 At 10,000 keys, avg. speed is 2,031 keys/sec
2492 At 20,000 keys, avg. speed is 1,970 keys/sec
2493 At 30,000 keys, avg. speed is 2,050 keys/sec
2494 At 40,000 keys, avg. speed is 2,073 keys/sec
2495 At 50,000 keys, avg. speed is 1,973 keys/sec
2496 At 60,000 keys, avg. speed is 1,914 keys/sec
2497 At 70,000 keys, avg. speed is 2,091 keys/sec
2498 At 80,000 keys, avg. speed is 2,103 keys/sec
2499 At 90,000 keys, avg. speed is 1,886 keys/sec
2500 At 100,000 keys, avg. speed is 1,970 keys/sec
2501 At 200,000 keys, avg. speed is 2,053 keys/sec
2502 At 300,000 keys, avg. speed is 1,697 keys/sec
2503 At 400,000 keys, avg. speed is 1,838 keys/sec
2504 At 500,000 keys, avg. speed is 1,941 keys/sec
2505 At 600,000 keys, avg. speed is 1,930 keys/sec
2506 At 700,000 keys, avg. speed is 1,735 keys/sec
2507 At 800,000 keys, avg. speed is 1,795 keys/sec
2508 At 900,000 keys, avg. speed is 1,221 keys/sec
2509 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2511 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2512 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2513 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2514 Run time was 12 min 3 sec.
2518 One of the great things about DBM::Deep is that it uses very little memory.
2519 Even with huge databases (1,000,000+ keys) you will not see much increased
2520 memory on your process. DBM::Deep relies solely on the filesystem for storing
2521 and fetching data. Here is output from I</usr/bin/top> before even opening a
2524 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2525 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2527 Basically the process is taking 2,716K of memory. And here is the same
2528 process after storing and fetching 1,000,000 keys:
2530 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2531 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2533 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2534 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2536 =head1 DB FILE FORMAT
2538 In case you were interested in the underlying DB file format, it is documented
2539 here in this section. You don't need to know this to use the module, it's just
2540 included for reference.
2544 DBM::Deep files always start with a 32-bit signature to identify the file type.
2545 This is at offset 0. The signature is "DPDB" in network byte order. This is
2546 checked for when the file is opened and an error will be thrown if it's not found.
2550 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2551 has a standard header containing the type of data, the length of data, and then
2552 the data itself. The type is a single character (1 byte), the length is a
2553 32-bit unsigned long in network byte order, and the data is, well, the data.
2554 Here is how it unfolds:
2558 Immediately after the 32-bit file signature is the I<Master Index> record.
2559 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2560 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2561 depending on how the DBM::Deep object was constructed.
2563 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2564 number). The first 8-bit char of the MD5 signature is the offset into the
2565 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2566 index element is a file offset of the next tag for the key/element in question,
2567 which is usually a I<Bucket List> tag (see below).
2569 The next tag I<could> be another index, depending on how many keys/elements
2570 exist. See L<RE-INDEXING> below for details.
2574 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2575 file offsets to where the actual data is stored. It starts with a standard
2576 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2577 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2578 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2579 When the list fills up, a I<Re-Index> operation is performed (See
2580 L<RE-INDEXING> below).
2584 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2585 index/value pair (in array mode). It starts with a standard tag header with
2586 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2587 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2588 header. The size reported in the tag header is only for the value, but then,
2589 just after the value is another size (32-bit unsigned long) and then the plain
2590 key itself. Since the value is likely to be fetched more often than the plain
2591 key, I figured it would be I<slightly> faster to store the value first.
2593 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2594 record for the nested structure, where the process begins all over again.
2598 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2599 exhausted. Then, when another key/element comes in, the list is converted to a
2600 new index record. However, this index will look at the next char in the MD5
2601 hash, and arrange new Bucket List pointers accordingly. This process is called
2602 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2603 17 (16 + new one) keys/elements are removed from the old Bucket List and
2604 inserted into the new index. Several new Bucket Lists are created in the
2605 process, as a new MD5 char from the key is being examined (it is unlikely that
2606 the keys will all share the same next char of their MD5s).
2608 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2609 when the Bucket Lists will turn into indexes, but the first round tends to
2610 happen right around 4,000 keys. You will see a I<slight> decrease in
2611 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2612 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2613 right around 900,000 keys. This process can continue nearly indefinitely --
2614 right up until the point the I<MD5> signatures start colliding with each other,
2615 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2616 getting struck by lightning while you are walking to cash in your tickets.
2617 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2618 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2619 this is 340 unodecillion, but don't quote me).
2623 When a new key/element is stored, the key (or index number) is first run through
2624 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2625 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2626 for the first char of the signature (in this case I<b0>). If it does not exist,
2627 a new I<Bucket List> is created for our key (and the next 15 future keys that
2628 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2629 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2630 this point, unless we are replacing an existing I<Bucket>), where the actual
2631 data will be stored.
2635 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2636 (or index number), then walking along the indexes. If there are enough
2637 keys/elements in this DB level, there might be nested indexes, each linked to
2638 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2639 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2640 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2641 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 We use B<Devel::Cover> to test the code coverage of our tests, below is the
2656 B<Devel::Cover> report on this module's test suite.
2658 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2659 File stmt bran cond sub pod time total
2660 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2661 blib/lib/DBM/Deep.pm 93.9 82.5 70.0 96.5 33.3 84.3 88.1
2662 blib/lib/DBM/Deep/Array.pm 98.8 88.9 87.5 100.0 n/a 9.0 96.4
2663 blib/lib/DBM/Deep/Hash.pm 95.2 80.0 100.0 100.0 n/a 6.7 92.3
2664 Total 95.0 83.4 73.8 97.6 33.3 100.0 89.9
2665 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2667 =head1 MORE INFORMATION
2669 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
2670 or send email to L<DBM-Deep@googlegroups.com>.
2674 Joseph Huckaby, L<jhuckaby@cpan.org>
2676 Rob Kinyon, L<rkinyon@cpan.org>
2678 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2682 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2683 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2687 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
2688 This is free software, you may use it and distribute it under the
2689 same terms as Perl itself.