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 $unpacked_key = $key;
1259 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1261 my $md5 = $DIGEST_FUNC->($key);
1264 # Make sure file is open
1266 if (!defined($self->fh) && !$self->_open()) {
1274 # Request exclusive lock for writing
1276 $self->lock( LOCK_EX );
1279 # If locking is enabled, set 'end' parameter again, in case another
1280 # DB instance appended to our file while we were unlocked.
1282 if ($self->root->{locking} || $self->root->{volatile}) {
1283 $self->root->{end} = (stat($fh))[7];
1287 # Locate offset for bucket list using digest index system
1289 my $tag = $self->_load_tag($self->base_offset);
1291 $tag = $self->_create_tag($self->base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1295 while ($tag->{signature} ne SIG_BLIST) {
1296 my $num = ord(substr($md5, $ch, 1));
1297 my $new_tag = $self->_index_lookup($tag, $num);
1299 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1300 seek($fh, $ref_loc, SEEK_SET);
1301 print($fh pack($LONG_PACK, $self->root->{end}) );
1303 $tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1304 $tag->{ref_loc} = $ref_loc;
1309 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1311 $tag->{ref_loc} = $ref_loc;
1318 # Add key/value to bucket list
1320 my $result = $self->_add_bucket( $tag, $md5, $key, $value );
1323 # If this object is an array, and bucket was not a replace, and key is numerical,
1324 # and index is equal or greater than current length, advance length variable.
1326 if (($result == 2) && ($self->type eq TYPE_ARRAY) && ($unpacked_key =~ /^\d+$/) && ($unpacked_key >= $self->FETCHSIZE())) {
1327 $self->STORESIZE( $unpacked_key + 1 );
1337 # Fetch single value or element given plain key or array index
1339 my $self = $_[0]->_get_self;
1342 if ( $self->type eq TYPE_HASH ) {
1343 if ( my $filter = $self->root->{filter_store_key} ) {
1344 $key = $filter->( $key );
1348 my $md5 = $DIGEST_FUNC->($key);
1351 # Make sure file is open
1353 if (!defined($self->fh)) { $self->_open(); }
1356 # Request shared lock for reading
1358 $self->lock( LOCK_SH );
1360 my $tag = $self->_find_bucket_list( $md5 );
1367 # Get value from bucket list
1369 my $result = $self->_get_bucket_value( $tag, $md5 );
1373 #XXX What is ref() checking here?
1374 return ($result && !ref($result) && $self->root->{filter_fetch_value}) ? $self->root->{filter_fetch_value}->($result) : $result;
1379 # Delete single key/value pair or element given plain key or array index
1381 my $self = $_[0]->_get_self;
1382 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1384 my $unpacked_key = $key;
1385 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1386 my $md5 = $DIGEST_FUNC->($key);
1389 # Make sure file is open
1391 if (!defined($self->fh)) { $self->_open(); }
1394 # Request exclusive lock for writing
1396 $self->lock( LOCK_EX );
1398 my $tag = $self->_find_bucket_list( $md5 );
1407 my $value = $self->FETCH( $unpacked_key );
1408 my $result = $self->_delete_bucket( $tag, $md5 );
1411 # If this object is an array and the key deleted was on the end of the stack,
1412 # decrement the length variable.
1414 if ($result && ($self->type eq TYPE_ARRAY) && ($unpacked_key == $self->FETCHSIZE() - 1)) {
1415 $self->STORESIZE( $unpacked_key );
1425 # Check if a single key or element exists given plain key or array index
1427 my $self = $_[0]->_get_self;
1428 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1430 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1431 my $md5 = $DIGEST_FUNC->($key);
1434 # Make sure file is open
1436 if (!defined($self->fh)) { $self->_open(); }
1439 # Request shared lock for reading
1441 $self->lock( LOCK_SH );
1443 my $tag = $self->_find_bucket_list( $md5 );
1446 # For some reason, the built-in exists() function returns '' for false
1454 # Check if bucket exists and return 1 or ''
1456 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1465 # Clear all keys from hash, or all elements from array.
1467 my $self = $_[0]->_get_self;
1470 # Make sure file is open
1472 if (!defined($self->fh)) { $self->_open(); }
1475 # Request exclusive lock for writing
1477 $self->lock( LOCK_EX );
1481 seek($fh, $self->base_offset, SEEK_SET);
1487 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
1495 # Public method aliases
1497 sub put { (shift)->STORE( @_ ) }
1498 sub store { (shift)->STORE( @_ ) }
1499 sub get { (shift)->FETCH( @_ ) }
1500 sub fetch { (shift)->FETCH( @_ ) }
1505 package DBM::Deep::_::Root;
1520 filter_store_key => undef,
1521 filter_store_value => undef,
1522 filter_fetch_key => undef,
1523 filter_fetch_value => undef,
1534 return unless $self;
1536 close $self->{fh} if $self->{fh};
1547 DBM::Deep - A pure perl multi-level hash/array DBM
1552 my $db = DBM::Deep->new( "foo.db" );
1554 $db->{key} = 'value'; # tie() style
1557 $db->put('key', 'value'); # OO style
1558 print $db->get('key');
1560 # true multi-level support
1561 $db->{my_complex} = [
1562 'hello', { perl => 'rules' },
1567 A unique flat-file database module, written in pure perl. True
1568 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1569 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1570 handle millions of keys and unlimited hash levels without significant
1571 slow-down. Written from the ground-up in pure perl -- this is NOT a
1572 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1573 Mac OS X and Windows.
1577 Hopefully you are using CPAN's excellent Perl module, which will download
1578 and install the module for you. If not, get the tarball, and run these
1590 Construction can be done OO-style (which is the recommended way), or using
1591 Perl's tie() function. Both are examined here.
1593 =head2 OO CONSTRUCTION
1595 The recommended way to construct a DBM::Deep object is to use the new()
1596 method, which gets you a blessed, tied hash or array reference.
1598 my $db = DBM::Deep->new( "foo.db" );
1600 This opens a new database handle, mapped to the file "foo.db". If this
1601 file does not exist, it will automatically be created. DB files are
1602 opened in "r+" (read/write) mode, and the type of object returned is a
1603 hash, unless otherwise specified (see L<OPTIONS> below).
1607 You can pass a number of options to the constructor to specify things like
1608 locking, autoflush, etc. This is done by passing an inline hash:
1610 my $db = DBM::Deep->new(
1616 Notice that the filename is now specified I<inside> the hash with
1617 the "file" parameter, as opposed to being the sole argument to the
1618 constructor. This is required if any options are specified.
1619 See L<OPTIONS> below for the complete list.
1623 You can also start with an array instead of a hash. For this, you must
1624 specify the C<type> parameter:
1626 my $db = DBM::Deep->new(
1628 type => DBM::Deep->TYPE_ARRAY
1631 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1632 a new DB file. If you create a DBM::Deep object with an existing file, the
1633 C<type> will be loaded from the file header, and ignored if it is passed
1636 =head2 TIE CONSTRUCTION
1638 Alternatively, you can create a DBM::Deep handle by using Perl's built-in
1639 tie() function. This is not ideal, because you get only a basic, tied hash
1640 (or array) which is not blessed, so you can't call any functions on it.
1643 tie %hash, "DBM::Deep", "foo.db";
1646 tie @array, "DBM::Deep", "bar.db";
1648 As with the OO constructor, you can replace the DB filename parameter with
1649 a hash containing one or more options (see L<OPTIONS> just below for the
1652 tie %hash, "DBM::Deep", {
1660 There are a number of options that can be passed in when constructing your
1661 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1667 Filename of the DB file to link the handle to. You can pass a full absolute
1668 filesystem path, partial path, or a plain filename if the file is in the
1669 current working directory. This is a required parameter.
1673 File open mode (read-only, read-write, etc.) string passed to Perl's FileHandle
1674 module. This is an optional parameter, and defaults to "r+" (read/write).
1675 B<Note:> If the default (r+) mode is selected, the file will also be auto-
1676 created if it doesn't exist.
1680 This parameter specifies what type of object to create, a hash or array. Use
1681 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1682 This only takes effect when beginning a new file. This is an optional
1683 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1687 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1688 function to lock the database in exclusive mode for writes, and shared mode for
1689 reads. Pass any true value to enable. This affects the base DB handle I<and
1690 any child hashes or arrays> that use the same DB file. This is an optional
1691 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1695 Specifies whether autoflush is to be enabled on the underlying FileHandle.
1696 This obviously slows down write operations, but is required if you may have
1697 multiple processes accessing the same DB file (also consider enable I<locking>
1698 or at least I<volatile>). Pass any true value to enable. This is an optional
1699 parameter, and defaults to 0 (disabled).
1703 If I<volatile> mode is enabled, DBM::Deep will stat() the DB file before each
1704 STORE() operation. This is required if an outside force may change the size of
1705 the file between transactions. Locking also implicitly enables volatile. This
1706 is useful if you want to use a different locking system or write your own. Pass
1707 any true value to enable. This is an optional parameter, and defaults to 0
1712 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1713 restore them when fetched. This is an B<experimental> feature, and does have
1714 side-effects. Basically, when hashes are re-blessed into their original
1715 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1716 able to call any DBM::Deep methods on them. You have been warned.
1717 This is an optional parameter, and defaults to 0 (disabled).
1721 See L<FILTERS> below.
1725 Setting I<debug> mode will make all errors non-fatal, dump them out to
1726 STDERR, and continue on. This is for debugging purposes only, and probably
1727 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1731 Instead of passing a file path, you can instead pass a handle to an pre-opened
1732 filehandle. Note: Beware of using the magick *DATA handle, as this actually
1733 contains your entire Perl script, as well as the data following the __DATA__
1734 marker. This will not work, because DBM::Deep uses absolute seek()s into the
1735 file. Instead, consider reading *DATA into an IO::Scalar handle, then passing
1740 =head1 TIE INTERFACE
1742 With DBM::Deep you can access your databases using Perl's standard hash/array
1743 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can treat
1744 them as such. DBM::Deep will intercept all reads/writes and direct them to the right
1745 place -- the DB file. This has nothing to do with the L<TIE CONSTRUCTION>
1746 section above. This simply tells you how to use DBM::Deep using regular hashes
1747 and arrays, rather than calling functions like C<get()> and C<put()> (although those
1748 work too). It is entirely up to you how to want to access your databases.
1752 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1753 or even nested hashes (or arrays) using standard Perl syntax:
1755 my $db = DBM::Deep->new( "foo.db" );
1757 $db->{mykey} = "myvalue";
1759 $db->{myhash}->{subkey} = "subvalue";
1761 print $db->{myhash}->{subkey} . "\n";
1763 You can even step through hash keys using the normal Perl C<keys()> function:
1765 foreach my $key (keys %$db) {
1766 print "$key: " . $db->{$key} . "\n";
1769 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1770 pushes them onto an array, all before the loop even begins. If you have an
1771 extra large hash, this may exhaust Perl's memory. Instead, consider using
1772 Perl's C<each()> function, which pulls keys/values one at a time, using very
1775 while (my ($key, $value) = each %$db) {
1776 print "$key: $value\n";
1779 Please note that when using C<each()>, you should always pass a direct
1780 hash reference, not a lookup. Meaning, you should B<never> do this:
1783 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1785 This causes an infinite loop, because for each iteration, Perl is calling
1786 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1787 it effectively keeps returning the first key over and over again. Instead,
1788 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1792 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1793 reference. This includes inserting, removing and manipulating elements,
1794 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1795 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1796 or simply be a nested array reference inside a hash. Example:
1798 my $db = DBM::Deep->new(
1799 file => "foo-array.db",
1800 type => DBM::Deep->TYPE_ARRAY
1804 push @$db, "bar", "baz";
1805 unshift @$db, "bah";
1807 my $last_elem = pop @$db; # baz
1808 my $first_elem = shift @$db; # bah
1809 my $second_elem = $db->[1]; # bar
1811 my $num_elements = scalar @$db;
1815 In addition to the I<tie()> interface, you can also use a standard OO interface
1816 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1817 array) has its own methods, but both types share the following common methods:
1818 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1824 Stores a new hash key/value pair, or sets an array element value. Takes two
1825 arguments, the hash key or array index, and the new value. The value can be
1826 a scalar, hash ref or array ref. Returns true on success, false on failure.
1828 $db->put("foo", "bar"); # for hashes
1829 $db->put(1, "bar"); # for arrays
1833 Fetches the value of a hash key or array element. Takes one argument: the hash
1834 key or array index. Returns a scalar, hash ref or array ref, depending on the
1837 my $value = $db->get("foo"); # for hashes
1838 my $value = $db->get(1); # for arrays
1842 Checks if a hash key or array index exists. Takes one argument: the hash key
1843 or array index. Returns true if it exists, false if not.
1845 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1846 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1850 Deletes one hash key/value pair or array element. Takes one argument: the hash
1851 key or array index. Returns true on success, false if not found. For arrays,
1852 the remaining elements located after the deleted element are NOT moved over.
1853 The deleted element is essentially just undefined, which is exactly how Perl's
1854 internal arrays work. Please note that the space occupied by the deleted
1855 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1856 below for details and workarounds.
1858 $db->delete("foo"); # for hashes
1859 $db->delete(1); # for arrays
1863 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1864 value. Please note that the space occupied by the deleted keys/values or
1865 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1866 details and workarounds.
1868 $db->clear(); # hashes or arrays
1874 For hashes, DBM::Deep supports all the common methods described above, and the
1875 following additional methods: C<first_key()> and C<next_key()>.
1881 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1882 fetched in an undefined order (which appears random). Takes no arguments,
1883 returns the key as a scalar value.
1885 my $key = $db->first_key();
1889 Returns the "next" key in the hash, given the previous one as the sole argument.
1890 Returns undef if there are no more keys to be fetched.
1892 $key = $db->next_key($key);
1896 Here are some examples of using hashes:
1898 my $db = DBM::Deep->new( "foo.db" );
1900 $db->put("foo", "bar");
1901 print "foo: " . $db->get("foo") . "\n";
1903 $db->put("baz", {}); # new child hash ref
1904 $db->get("baz")->put("buz", "biz");
1905 print "buz: " . $db->get("baz")->get("buz") . "\n";
1907 my $key = $db->first_key();
1909 print "$key: " . $db->get($key) . "\n";
1910 $key = $db->next_key($key);
1913 if ($db->exists("foo")) { $db->delete("foo"); }
1917 For arrays, DBM::Deep supports all the common methods described above, and the
1918 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1919 C<unshift()> and C<splice()>.
1925 Returns the number of elements in the array. Takes no arguments.
1927 my $len = $db->length();
1931 Adds one or more elements onto the end of the array. Accepts scalars, hash
1932 refs or array refs. No return value.
1934 $db->push("foo", "bar", {});
1938 Fetches the last element in the array, and deletes it. Takes no arguments.
1939 Returns undef if array is empty. Returns the element value.
1941 my $elem = $db->pop();
1945 Fetches the first element in the array, deletes it, then shifts all the
1946 remaining elements over to take up the space. Returns the element value. This
1947 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1950 my $elem = $db->shift();
1954 Inserts one or more elements onto the beginning of the array, shifting all
1955 existing elements over to make room. Accepts scalars, hash refs or array refs.
1956 No return value. This method is not recommended with large arrays -- see
1957 <LARGE ARRAYS> below for details.
1959 $db->unshift("foo", "bar", {});
1963 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1964 -f splice> for usage -- it is too complicated to document here. This method is
1965 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1969 Here are some examples of using arrays:
1971 my $db = DBM::Deep->new(
1973 type => DBM::Deep->TYPE_ARRAY
1976 $db->push("bar", "baz");
1977 $db->unshift("foo");
1980 my $len = $db->length();
1981 print "length: $len\n"; # 4
1983 for (my $k=0; $k<$len; $k++) {
1984 print "$k: " . $db->get($k) . "\n";
1987 $db->splice(1, 2, "biz", "baf");
1989 while (my $elem = shift @$db) {
1990 print "shifted: $elem\n";
1995 Enable automatic file locking by passing a true value to the C<locking>
1996 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1998 my $db = DBM::Deep->new(
2003 This causes DBM::Deep to C<flock()> the underlying FileHandle object with exclusive
2004 mode for writes, and shared mode for reads. This is required if you have
2005 multiple processes accessing the same database file, to avoid file corruption.
2006 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
2007 NFS> below for more.
2009 =head2 EXPLICIT LOCKING
2011 You can explicitly lock a database, so it remains locked for multiple
2012 transactions. This is done by calling the C<lock()> method, and passing an
2013 optional lock mode argument (defaults to exclusive mode). This is particularly
2014 useful for things like counters, where the current value needs to be fetched,
2015 then incremented, then stored again.
2018 my $counter = $db->get("counter");
2020 $db->put("counter", $counter);
2029 You can pass C<lock()> an optional argument, which specifies which mode to use
2030 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2031 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2032 same as the constants defined in Perl's C<Fcntl> module.
2034 $db->lock( DBM::Deep->LOCK_SH );
2038 If you want to implement your own file locking scheme, be sure to create your
2039 DBM::Deep objects setting the C<volatile> option to true. This hints to DBM::Deep
2040 that the DB file may change between transactions. See L<LOW-LEVEL ACCESS>
2043 =head1 IMPORTING/EXPORTING
2045 You can import existing complex structures by calling the C<import()> method,
2046 and export an entire database into an in-memory structure using the C<export()>
2047 method. Both are examined here.
2051 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2052 walking the structure and adding keys/elements to the database as you go,
2053 simply pass a reference to the C<import()> method. This recursively adds
2054 everything to an existing DBM::Deep object for you. Here is an example:
2059 array1 => [ "elem0", "elem1", "elem2" ],
2061 subkey1 => "subvalue1",
2062 subkey2 => "subvalue2"
2066 my $db = DBM::Deep->new( "foo.db" );
2067 $db->import( $struct );
2069 print $db->{key1} . "\n"; # prints "value1"
2071 This recursively imports the entire C<$struct> object into C<$db>, including
2072 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2073 keys are merged with the existing ones, replacing if they already exist.
2074 The C<import()> method can be called on any database level (not just the base
2075 level), and works with both hash and array DB types.
2079 B<Note:> Make sure your existing structure has no circular references in it.
2080 These will cause an infinite loop when importing.
2084 Calling the C<export()> method on an existing DBM::Deep object will return
2085 a reference to a new in-memory copy of the database. The export is done
2086 recursively, so all nested hashes/arrays are all exported to standard Perl
2087 objects. Here is an example:
2089 my $db = DBM::Deep->new( "foo.db" );
2091 $db->{key1} = "value1";
2092 $db->{key2} = "value2";
2094 $db->{hash1}->{subkey1} = "subvalue1";
2095 $db->{hash1}->{subkey2} = "subvalue2";
2097 my $struct = $db->export();
2099 print $struct->{key1} . "\n"; # prints "value1"
2101 This makes a complete copy of the database in memory, and returns a reference
2102 to it. The C<export()> method can be called on any database level (not just
2103 the base level), and works with both hash and array DB types. Be careful of
2104 large databases -- you can store a lot more data in a DBM::Deep object than an
2105 in-memory Perl structure.
2109 B<Note:> Make sure your database has no circular references in it.
2110 These will cause an infinite loop when exporting.
2114 DBM::Deep has a number of hooks where you can specify your own Perl function
2115 to perform filtering on incoming or outgoing data. This is a perfect
2116 way to extend the engine, and implement things like real-time compression or
2117 encryption. Filtering applies to the base DB level, and all child hashes /
2118 arrays. Filter hooks can be specified when your DBM::Deep object is first
2119 constructed, or by calling the C<set_filter()> method at any time. There are
2120 four available filter hooks, described below:
2124 =item * filter_store_key
2126 This filter is called whenever a hash key is stored. It
2127 is passed the incoming key, and expected to return a transformed key.
2129 =item * filter_store_value
2131 This filter is called whenever a hash key or array element is stored. It
2132 is passed the incoming value, and expected to return a transformed value.
2134 =item * filter_fetch_key
2136 This filter is called whenever a hash key is fetched (i.e. via
2137 C<first_key()> or C<next_key()>). It is passed the transformed key,
2138 and expected to return the plain key.
2140 =item * filter_fetch_value
2142 This filter is called whenever a hash key or array element is fetched.
2143 It is passed the transformed value, and expected to return the plain value.
2147 Here are the two ways to setup a filter hook:
2149 my $db = DBM::Deep->new(
2151 filter_store_value => \&my_filter_store,
2152 filter_fetch_value => \&my_filter_fetch
2157 $db->set_filter( "filter_store_value", \&my_filter_store );
2158 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2160 Your filter function will be called only when dealing with SCALAR keys or
2161 values. When nested hashes and arrays are being stored/fetched, filtering
2162 is bypassed. Filters are called as static functions, passed a single SCALAR
2163 argument, and expected to return a single SCALAR value. If you want to
2164 remove a filter, set the function reference to C<undef>:
2166 $db->set_filter( "filter_store_value", undef );
2168 =head2 REAL-TIME ENCRYPTION EXAMPLE
2170 Here is a working example that uses the I<Crypt::Blowfish> module to
2171 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2172 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2173 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2176 use Crypt::Blowfish;
2179 my $cipher = Crypt::CBC->new({
2180 'key' => 'my secret key',
2181 'cipher' => 'Blowfish',
2183 'regenerate_key' => 0,
2184 'padding' => 'space',
2188 my $db = DBM::Deep->new(
2189 file => "foo-encrypt.db",
2190 filter_store_key => \&my_encrypt,
2191 filter_store_value => \&my_encrypt,
2192 filter_fetch_key => \&my_decrypt,
2193 filter_fetch_value => \&my_decrypt,
2196 $db->{key1} = "value1";
2197 $db->{key2} = "value2";
2198 print "key1: " . $db->{key1} . "\n";
2199 print "key2: " . $db->{key2} . "\n";
2205 return $cipher->encrypt( $_[0] );
2208 return $cipher->decrypt( $_[0] );
2211 =head2 REAL-TIME COMPRESSION EXAMPLE
2213 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2214 compression / decompression of keys & values with DBM::Deep Filters.
2215 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2216 more on I<Compress::Zlib>.
2221 my $db = DBM::Deep->new(
2222 file => "foo-compress.db",
2223 filter_store_key => \&my_compress,
2224 filter_store_value => \&my_compress,
2225 filter_fetch_key => \&my_decompress,
2226 filter_fetch_value => \&my_decompress,
2229 $db->{key1} = "value1";
2230 $db->{key2} = "value2";
2231 print "key1: " . $db->{key1} . "\n";
2232 print "key2: " . $db->{key2} . "\n";
2238 return Compress::Zlib::memGzip( $_[0] ) ;
2241 return Compress::Zlib::memGunzip( $_[0] ) ;
2244 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2245 actually numerical index numbers, and are not filtered.
2247 =head1 ERROR HANDLING
2249 Most DBM::Deep methods return a true value for success, and call die() on
2250 failure. You can wrap calls in an eval block to catch the die. Also, the
2251 actual error message is stored in an internal scalar, which can be fetched by
2252 calling the C<error()> method.
2254 my $db = DBM::Deep->new( "foo.db" ); # create hash
2255 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2257 print $db->error(); # prints error message
2259 You can then call C<clear_error()> to clear the current error state.
2263 If you set the C<debug> option to true when creating your DBM::Deep object,
2264 all errors are considered NON-FATAL, and dumped to STDERR. This is only
2265 for debugging purposes.
2267 =head1 LARGEFILE SUPPORT
2269 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2270 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2271 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2272 by calling the static C<set_pack()> method before you do anything else.
2274 DBM::Deep::set_pack(8, 'Q');
2276 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2277 instead of 32-bit longs. After setting these values your DB files have a
2278 theoretical maximum size of 16 XB (exabytes).
2282 B<Note:> Changing these values will B<NOT> work for existing database files.
2283 Only change this for new files, and make sure it stays set consistently
2284 throughout the file's life. If you do set these values, you can no longer
2285 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2286 back to 32-bit mode.
2290 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2291 only a 32-bit Perl. However, I have received user reports that this does
2294 =head1 LOW-LEVEL ACCESS
2296 If you require low-level access to the underlying FileHandle that DBM::Deep uses,
2297 you can call the C<fh()> method, which returns the handle:
2301 This method can be called on the root level of the datbase, or any child
2302 hashes or arrays. All levels share a I<root> structure, which contains things
2303 like the FileHandle, a reference counter, and all your options you specified
2304 when you created the object. You can get access to this root structure by
2305 calling the C<root()> method.
2307 my $root = $db->root();
2309 This is useful for changing options after the object has already been created,
2310 such as enabling/disabling locking, volatile or debug modes. You can also
2311 store your own temporary user data in this structure (be wary of name
2312 collision), which is then accessible from any child hash or array.
2314 =head1 CUSTOM DIGEST ALGORITHM
2316 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2317 keys. However you can override this, and use another algorithm (such as SHA-256)
2318 or even write your own. But please note that DBM::Deep currently expects zero
2319 collisions, so your algorithm has to be I<perfect>, so to speak.
2320 Collision detection may be introduced in a later version.
2324 You can specify a custom digest algorithm by calling the static C<set_digest()>
2325 function, passing a reference to a subroutine, and the length of the algorithm's
2326 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2327 objects. Here is a working example that uses a 256-bit hash from the
2328 I<Digest::SHA256> module. Please see
2329 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2334 my $context = Digest::SHA256::new(256);
2336 DBM::Deep::set_digest( \&my_digest, 32 );
2338 my $db = DBM::Deep->new( "foo-sha.db" );
2340 $db->{key1} = "value1";
2341 $db->{key2} = "value2";
2342 print "key1: " . $db->{key1} . "\n";
2343 print "key2: " . $db->{key2} . "\n";
2349 return substr( $context->hash($_[0]), 0, 32 );
2352 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2353 of bytes you specify in the C<set_digest()> function (in this case 32).
2355 =head1 CIRCULAR REFERENCES
2357 DBM::Deep has B<experimental> support for circular references. Meaning you
2358 can have a nested hash key or array element that points to a parent object.
2359 This relationship is stored in the DB file, and is preserved between sessions.
2362 my $db = DBM::Deep->new( "foo.db" );
2365 $db->{circle} = $db; # ref to self
2367 print $db->{foo} . "\n"; # prints "foo"
2368 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2370 One catch is, passing the object to a function that recursively walks the
2371 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2372 C<export()> methods) will result in an infinite loop. The other catch is,
2373 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2374 or C<next_key()> methods), you will get the I<target object's key>, not the
2375 ref's key. This gets even more interesting with the above example, where
2376 the I<circle> key points to the base DB object, which technically doesn't
2377 have a key. So I made DBM::Deep return "[base]" as the key name in that
2380 =head1 CAVEATS / ISSUES / BUGS
2382 This section describes all the known issues with DBM::Deep. It you have found
2383 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2385 =head2 UNUSED SPACE RECOVERY
2387 One major caveat with DBM::Deep is that space occupied by existing keys and
2388 values is not recovered when they are deleted. Meaning if you keep deleting
2389 and adding new keys, your file will continuously grow. I am working on this,
2390 but in the meantime you can call the built-in C<optimize()> method from time to
2391 time (perhaps in a crontab or something) to recover all your unused space.
2393 $db->optimize(); # returns true on success
2395 This rebuilds the ENTIRE database into a new file, then moves it on top of
2396 the original. The new file will have no unused space, thus it will take up as
2397 little disk space as possible. Please note that this operation can take
2398 a long time for large files, and you need enough disk space to temporarily hold
2399 2 copies of your DB file. The temporary file is created in the same directory
2400 as the original, named with a ".tmp" extension, and is deleted when the
2401 operation completes. Oh, and if locking is enabled, the DB is automatically
2402 locked for the entire duration of the copy.
2406 B<WARNING:> Only call optimize() on the top-level node of the database, and
2407 make sure there are no child references lying around. DBM::Deep keeps a reference
2408 counter, and if it is greater than 1, optimize() will abort and return undef.
2410 =head2 AUTOVIVIFICATION
2412 Unfortunately, autovivification doesn't work with tied hashes. This appears to
2413 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
2414 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
2415 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
2416 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
2419 $db->{foo}->{bar} = "hello";
2421 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
2422 being an empty hash. Try this instead, which works fine:
2424 $db->{foo} = { bar => "hello" };
2426 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
2427 the execution path, and Perl indeed passes an empty hash to the STORE() method.
2428 Probably a bug in Perl.
2430 =head2 FILE CORRUPTION
2432 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2433 for a 32-bit signature when opened, but other corruption in files can cause
2434 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2435 stuck in an infinite loop depending on the level of corruption. File write
2436 operations are not checked for failure (for speed), so if you happen to run
2437 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2438 be addressed in a later version of DBM::Deep.
2442 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2443 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2444 about setting up your NFS server with a locking daemon, then using lockf() to
2445 lock your files, but your milage may vary there as well. From what I
2446 understand, there is no real way to do it. However, if you need access to the
2447 underlying FileHandle in DBM::Deep for using some other kind of locking scheme like
2448 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2450 =head2 COPYING OBJECTS
2452 Beware of copying tied objects in Perl. Very strange things can happen.
2453 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2454 returns a new, blessed, tied hash or array to the same level in the DB.
2456 my $copy = $db->clone();
2460 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2461 These functions cause every element in the array to move, which can be murder
2462 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2463 a different location. This may be addressed in a later version.
2467 This section discusses DBM::Deep's speed and memory usage.
2471 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2472 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2473 multi-level hash/array support, and cross-platform FTPable files. Even so,
2474 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2475 with huge databases. Here is some test data:
2477 Adding 1,000,000 keys to new DB file...
2479 At 100 keys, avg. speed is 2,703 keys/sec
2480 At 200 keys, avg. speed is 2,642 keys/sec
2481 At 300 keys, avg. speed is 2,598 keys/sec
2482 At 400 keys, avg. speed is 2,578 keys/sec
2483 At 500 keys, avg. speed is 2,722 keys/sec
2484 At 600 keys, avg. speed is 2,628 keys/sec
2485 At 700 keys, avg. speed is 2,700 keys/sec
2486 At 800 keys, avg. speed is 2,607 keys/sec
2487 At 900 keys, avg. speed is 2,190 keys/sec
2488 At 1,000 keys, avg. speed is 2,570 keys/sec
2489 At 2,000 keys, avg. speed is 2,417 keys/sec
2490 At 3,000 keys, avg. speed is 1,982 keys/sec
2491 At 4,000 keys, avg. speed is 1,568 keys/sec
2492 At 5,000 keys, avg. speed is 1,533 keys/sec
2493 At 6,000 keys, avg. speed is 1,787 keys/sec
2494 At 7,000 keys, avg. speed is 1,977 keys/sec
2495 At 8,000 keys, avg. speed is 2,028 keys/sec
2496 At 9,000 keys, avg. speed is 2,077 keys/sec
2497 At 10,000 keys, avg. speed is 2,031 keys/sec
2498 At 20,000 keys, avg. speed is 1,970 keys/sec
2499 At 30,000 keys, avg. speed is 2,050 keys/sec
2500 At 40,000 keys, avg. speed is 2,073 keys/sec
2501 At 50,000 keys, avg. speed is 1,973 keys/sec
2502 At 60,000 keys, avg. speed is 1,914 keys/sec
2503 At 70,000 keys, avg. speed is 2,091 keys/sec
2504 At 80,000 keys, avg. speed is 2,103 keys/sec
2505 At 90,000 keys, avg. speed is 1,886 keys/sec
2506 At 100,000 keys, avg. speed is 1,970 keys/sec
2507 At 200,000 keys, avg. speed is 2,053 keys/sec
2508 At 300,000 keys, avg. speed is 1,697 keys/sec
2509 At 400,000 keys, avg. speed is 1,838 keys/sec
2510 At 500,000 keys, avg. speed is 1,941 keys/sec
2511 At 600,000 keys, avg. speed is 1,930 keys/sec
2512 At 700,000 keys, avg. speed is 1,735 keys/sec
2513 At 800,000 keys, avg. speed is 1,795 keys/sec
2514 At 900,000 keys, avg. speed is 1,221 keys/sec
2515 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2517 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2518 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2519 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2520 Run time was 12 min 3 sec.
2524 One of the great things about DBM::Deep is that it uses very little memory.
2525 Even with huge databases (1,000,000+ keys) you will not see much increased
2526 memory on your process. DBM::Deep relies solely on the filesystem for storing
2527 and fetching data. Here is output from I</usr/bin/top> before even opening a
2530 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2531 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2533 Basically the process is taking 2,716K of memory. And here is the same
2534 process after storing and fetching 1,000,000 keys:
2536 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2537 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2539 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2540 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2542 =head1 DB FILE FORMAT
2544 In case you were interested in the underlying DB file format, it is documented
2545 here in this section. You don't need to know this to use the module, it's just
2546 included for reference.
2550 DBM::Deep files always start with a 32-bit signature to identify the file type.
2551 This is at offset 0. The signature is "DPDB" in network byte order. This is
2552 checked when the file is opened.
2556 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2557 has a standard header containing the type of data, the length of data, and then
2558 the data itself. The type is a single character (1 byte), the length is a
2559 32-bit unsigned long in network byte order, and the data is, well, the data.
2560 Here is how it unfolds:
2564 Immediately after the 32-bit file signature is the I<Master Index> record.
2565 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2566 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2567 depending on how the DBM::Deep object was constructed.
2571 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2572 number). The first 8-bit char of the MD5 signature is the offset into the
2573 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2574 index element is a file offset of the next tag for the key/element in question,
2575 which is usually a I<Bucket List> tag (see below).
2579 The next tag I<could> be another index, depending on how many keys/elements
2580 exist. See L<RE-INDEXING> below for details.
2584 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2585 file offsets to where the actual data is stored. It starts with a standard
2586 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2587 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2588 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2589 When the list fills up, a I<Re-Index> operation is performed (See
2590 L<RE-INDEXING> below).
2594 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2595 index/value pair (in array mode). It starts with a standard tag header with
2596 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2597 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2598 header. The size reported in the tag header is only for the value, but then,
2599 just after the value is another size (32-bit unsigned long) and then the plain
2600 key itself. Since the value is likely to be fetched more often than the plain
2601 key, I figured it would be I<slightly> faster to store the value first.
2605 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2606 record for the nested structure, where the process begins all over again.
2610 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2611 exhausted. Then, when another key/element comes in, the list is converted to a
2612 new index record. However, this index will look at the next char in the MD5
2613 hash, and arrange new Bucket List pointers accordingly. This process is called
2614 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2615 17 (16 + new one) keys/elements are removed from the old Bucket List and
2616 inserted into the new index. Several new Bucket Lists are created in the
2617 process, as a new MD5 char from the key is being examined (it is unlikely that
2618 the keys will all share the same next char of their MD5s).
2622 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2623 when the Bucket Lists will turn into indexes, but the first round tends to
2624 happen right around 4,000 keys. You will see a I<slight> decrease in
2625 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2626 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2627 right around 900,000 keys. This process can continue nearly indefinitely --
2628 right up until the point the I<MD5> signatures start colliding with each other,
2629 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2630 getting struck by lightning while you are walking to cash in your tickets.
2631 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2632 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2633 this is 340 unodecillion, but don't quote me).
2637 When a new key/element is stored, the key (or index number) is first ran through
2638 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2639 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2640 for the first char of the signature (in this case I<b>). If it does not exist,
2641 a new I<Bucket List> is created for our key (and the next 15 future keys that
2642 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2643 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2644 this point, unless we are replacing an existing I<Bucket>), where the actual
2645 data will be stored.
2649 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2650 (or index number), then walking along the indexes. If there are enough
2651 keys/elements in this DB level, there might be nested indexes, each linked to
2652 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2653 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2654 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2655 plain key are stored.
2659 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2660 methods. In this process the indexes are walked systematically, and each key
2661 fetched in increasing MD5 order (which is why it appears random). Once the
2662 I<Bucket> is found, the value is skipped the plain key returned instead.
2663 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2664 alphabetically sorted. This only happens on an index-level -- as soon as the
2665 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2666 so it's pretty much undefined how the keys will come out -- just like Perl's
2669 =head1 CODE COVERAGE
2671 I use B<Devel::Cover> to test the code coverage of my tests, below is the B<Devel::Cover> report on this
2672 module's test suite.
2674 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2675 File stmt bran cond sub pod time total
2676 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2677 blib/lib/DBM/Deep.pm 93.9 82.4 74.7 97.9 10.5 85.7 88.0
2678 blib/lib/DBM/Deep/Array.pm 97.8 84.6 50.0 100.0 n/a 9.0 94.6
2679 blib/lib/DBM/Deep/Hash.pm 93.9 87.5 100.0 100.0 n/a 5.3 93.4
2680 Total 94.4 82.9 75.8 98.5 10.5 100.0 89.0
2681 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2685 Joseph Huckaby, L<jhuckaby@cpan.org>
2687 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2691 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2692 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2696 Copyright (c) 2002-2005 Joseph Huckaby. All Rights Reserved.
2697 This is free software, you may use it and distribute it under the
2698 same terms as Perl itself.