7 # Multi-level database module for storing hash trees, arrays and simple
8 # key/value pairs into FTP-able, cross-platform binary database files.
10 # Type `perldoc DBM::Deep` for complete documentation.
14 # tie %db, 'DBM::Deep', 'my_database.db'; # standard tie() method
16 # my $db = new DBM::Deep( 'my_database.db' ); # preferred OO method
18 # $db->{my_scalar} = 'hello world';
19 # $db->{my_hash} = { larry => 'genius', hashes => 'fast' };
20 # $db->{my_array} = [ 1, 2, 3, time() ];
21 # $db->{my_complex} = [ 'hello', { perl => 'rules' }, 42, 99 ];
22 # push @{$db->{my_array}}, 'another value';
23 # my @key_list = keys %{$db->{my_hash}};
24 # print "This module " . $db->{my_complex}->[1]->{perl} . "!\n";
27 # (c) 2002-2005 Joseph Huckaby. All Rights Reserved.
28 # This program is free software; you can redistribute it and/or
29 # modify it under the same terms as Perl itself.
34 use Fcntl qw( :DEFAULT :flock :seek );
38 use vars qw( $VERSION );
42 # Set to 4 and 'N' for 32-bit offset tags (default). Theoretical limit of 4 GB per file.
43 # (Perl must be compiled with largefile support for files > 2 GB)
45 # Set to 8 and 'Q' for 64-bit offsets. Theoretical limit of 16 XB per file.
46 # (Perl must be compiled with largefile and 64-bit long support)
52 # Set to 4 and 'N' for 32-bit data length prefixes. Limit of 4 GB for each key/value.
53 # Upgrading this is possible (see above) but probably not necessary. If you need
54 # more than 4 GB for a single key or value, this module is really not for you :-)
56 #my $DATA_LENGTH_SIZE = 4;
57 #my $DATA_LENGTH_PACK = 'N';
58 our ($LONG_SIZE, $LONG_PACK, $DATA_LENGTH_SIZE, $DATA_LENGTH_PACK);
61 # Maximum number of buckets per list before another level of indexing is done.
62 # Increase this value for slightly greater speed, but larger database files.
63 # DO NOT decrease this value below 16, due to risk of recursive reindex overrun.
68 # Better not adjust anything below here, unless you're me :-)
72 # Setup digest function for keys
74 our ($DIGEST_FUNC, $HASH_SIZE);
75 #my $DIGEST_FUNC = \&Digest::MD5::md5;
78 # Precalculate index and bucket sizes based on values above.
81 my ($INDEX_SIZE, $BUCKET_SIZE, $BUCKET_LIST_SIZE);
88 # Setup file and tag signatures. These should never change.
90 sub SIG_FILE () { 'DPDB' }
91 sub SIG_HASH () { 'H' }
92 sub SIG_ARRAY () { 'A' }
93 sub SIG_SCALAR () { 'S' }
94 sub SIG_NULL () { 'N' }
95 sub SIG_DATA () { 'D' }
96 sub SIG_INDEX () { 'I' }
97 sub SIG_BLIST () { 'B' }
101 # Setup constants for users to pass to new()
103 sub TYPE_HASH () { return SIG_HASH; }
104 sub TYPE_ARRAY () { return SIG_ARRAY; }
105 sub TYPE_SCALAR () { return SIG_SCALAR; }
111 if (scalar(@_) > 1) {
113 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
117 elsif ( my $type = Scalar::Util::reftype($_[0]) ) {
118 if ( $type ne 'HASH' ) {
119 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
124 $args = { file => shift };
132 # Class constructor method for Perl OO interface.
133 # Calls tie() and returns blessed reference to tied hash or array,
134 # providing a hybrid OO/tie interface.
137 my $args = $class->_get_args( @_ );
140 # Check if we want a tied hash or array.
143 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
144 $class = 'DBM::Deep::Array';
145 require DBM::Deep::Array;
146 tie @$self, $class, %$args;
149 $class = 'DBM::Deep::Hash';
150 require DBM::Deep::Hash;
151 tie %$self, $class, %$args;
154 return bless $self, $class;
159 # Setup $self and bless into this class.
164 # These are the defaults to be optionally overridden below
167 base_offset => length(SIG_FILE),
170 foreach my $param ( keys %$self ) {
171 next unless exists $args->{$param};
172 $self->{$param} = delete $args->{$param}
175 $self->{root} = exists $args->{root}
177 : DBM::Deep::_::Root->new( $args );
179 if (!defined($self->fh)) { $self->_open(); }
186 require DBM::Deep::Hash;
187 return DBM::Deep::Hash->TIEHASH( @_ );
192 require DBM::Deep::Array;
193 return DBM::Deep::Array->TIEARRAY( @_ );
196 #XXX Unneeded now ...
202 # Open a FileHandle to the database, create if nonexistent.
203 # Make sure file signature matches DeepDB spec.
205 my $self = $_[0]->_get_self;
207 if (defined($self->fh)) { $self->_close(); }
210 # Theoretically, adding O_BINARY should remove the need for the binmode
211 # Of course, testing it is going to be ... interesting.
212 my $flags = O_RDWR | O_CREAT | O_BINARY;
214 #XXX Can the mode be anything but r+, w+, or a+??
215 #XXX ie, it has to be in read-write mode
216 #XXX So, should we verify that the mode is legitimate?
218 #XXX Maybe the mode thingy should just go away. There's no good
219 #XXX reason for it ...
220 if ( $self->root->{mode} eq 'w+' ) {
225 sysopen( $fh, $self->root->{file}, $flags )
227 $self->root->{fh} = $fh;
228 }; if ($@ ) { $self->_throw_error( "Received error: $@\n" ); }
229 if (! defined($self->fh)) {
230 return $self->_throw_error("Cannot sysopen file: " . $self->root->{file} . ": $!");
235 #XXX Can we remove this by using the right sysopen() flags?
236 # Maybe ... q.v. above
237 binmode $fh; # for win32
239 if ($self->root->{autoflush}) {
240 my $old = select $fh;
246 seek($fh, 0, SEEK_SET);
249 my $bytes_read = read( $fh, $signature, length(SIG_FILE));
252 # File is empty -- write signature and master index
255 seek($fh, 0, SEEK_SET);
257 $self->root->{end} = length(SIG_FILE);
258 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
260 my $plain_key = "[base]";
261 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
262 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
264 # Flush the filehandle
265 my $old_fh = select $fh;
275 # Check signature was valid
277 unless ($signature eq SIG_FILE) {
279 return $self->_throw_error("Signature not found -- file is not a Deep DB");
282 $self->root->{end} = (stat($fh))[7];
285 # Get our type from master index signature
287 my $tag = $self->_load_tag($self->base_offset);
289 #XXX We probably also want to store the hash algorithm name and not assume anything
290 #XXX The cool thing would be to allow a different hashing algorithm at every level
293 return $self->_throw_error("Corrupted file, no master index record");
295 if ($self->{type} ne $tag->{signature}) {
296 return $self->_throw_error("File type mismatch");
304 # Close database FileHandle
306 my $self = $_[0]->_get_self;
307 close $self->root->{fh};
308 $self->root->{fh} = undef;
313 # Given offset, signature and content, create tag and write to disk
315 my ($self, $offset, $sig, $content) = @_;
316 my $size = length($content);
320 seek($fh, $offset, SEEK_SET);
321 print($fh $sig . pack($DATA_LENGTH_PACK, $size) . $content );
323 if ($offset == $self->root->{end}) {
324 $self->root->{end} += SIG_SIZE + $DATA_LENGTH_SIZE + $size;
330 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
337 # Given offset, load single tag and return signature, size and data
344 seek($fh, $offset, SEEK_SET);
345 if (eof $fh) { return undef; }
348 read( $fh, $sig, SIG_SIZE);
351 read( $fh, $size, $DATA_LENGTH_SIZE);
352 $size = unpack($DATA_LENGTH_PACK, $size);
355 read( $fh, $buffer, $size);
360 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
367 # Given index tag, lookup single entry in index and return .
370 my ($tag, $index) = @_;
372 my $location = unpack($LONG_PACK, substr($tag->{content}, $index * $LONG_SIZE, $LONG_SIZE) );
373 if (!$location) { return; }
375 return $self->_load_tag( $location );
380 # Adds one key/value pair to bucket list, given offset, MD5 digest of key,
381 # plain (undigested) key and value.
384 my ($tag, $md5, $plain_key, $value) = @_;
385 my $keys = $tag->{content};
389 my $is_dbm_deep = 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' ) { $actual_length = $INDEX_SIZE; }
438 else { $actual_length = length($value); }
440 if ($actual_length <= $size) {
444 $location = $self->root->{end};
445 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE, SEEK_SET);
446 print($fh pack($LONG_PACK, $location) );
454 # If this is an internal reference, return now.
455 # No need to write value or plain key
462 # If bucket didn't fit into list, split into a new index level
465 seek($fh, $tag->{ref_loc}, SEEK_SET);
466 print($fh pack($LONG_PACK, $self->root->{end}) );
468 my $index_tag = $self->_create_tag($self->root->{end}, SIG_INDEX, chr(0) x $INDEX_SIZE);
471 $keys .= $md5 . pack($LONG_PACK, 0);
473 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
474 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
476 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
477 my $num = ord(substr($key, $tag->{ch} + 1, 1));
479 if ($offsets[$num]) {
480 my $offset = $offsets[$num] + SIG_SIZE + $DATA_LENGTH_SIZE;
481 seek($fh, $offset, SEEK_SET);
483 read( $fh, $subkeys, $BUCKET_LIST_SIZE);
485 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
486 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
488 seek($fh, $offset + ($k * $BUCKET_SIZE), SEEK_SET);
489 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
495 $offsets[$num] = $self->root->{end};
496 seek($fh, $index_tag->{offset} + ($num * $LONG_SIZE), SEEK_SET);
497 print($fh pack($LONG_PACK, $self->root->{end}) );
499 my $blist_tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
501 seek($fh, $blist_tag->{offset}, SEEK_SET);
502 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
507 $location ||= $self->root->{end};
508 } # re-index bucket list
511 # Seek to content area and store signature, value and plaintext key
515 seek($fh, $location, SEEK_SET);
518 # Write signature based on content type, set content length and write actual value.
520 my $r = Scalar::Util::reftype($value) || '';
522 print($fh TYPE_HASH );
523 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
524 $content_length = $INDEX_SIZE;
526 elsif ($r eq 'ARRAY') {
527 print($fh TYPE_ARRAY );
528 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
529 $content_length = $INDEX_SIZE;
531 elsif (!defined($value)) {
532 print($fh SIG_NULL );
533 print($fh pack($DATA_LENGTH_PACK, 0) );
537 print($fh SIG_DATA );
538 print($fh pack($DATA_LENGTH_PACK, length($value)) . $value );
539 $content_length = length($value);
543 # Plain key is stored AFTER value, as keys are typically fetched less often.
545 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
548 # If value is blessed, preserve class name
550 if ( $self->root->{autobless} ) {
551 my $value_class = Scalar::Util::blessed($value);
552 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
554 # Blessed ref -- will restore later
557 print($fh pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
558 $content_length += 1;
559 $content_length += $DATA_LENGTH_SIZE + length($value_class);
563 $content_length += 1;
568 # If this is a new content area, advance EOF counter
570 if ($location == $self->root->{end}) {
571 $self->root->{end} += SIG_SIZE;
572 $self->root->{end} += $DATA_LENGTH_SIZE + $content_length;
573 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
577 # If content is a hash or array, create new child DeepDB object and
578 # pass each key or element to it.
581 my $branch = DBM::Deep->new(
583 base_offset => $location,
586 foreach my $key (keys %{$value}) {
587 #$branch->{$key} = $value->{$key};
588 $branch->STORE( $key, $value->{$key} );
591 elsif ($r eq 'ARRAY') {
592 my $branch = DBM::Deep->new(
594 base_offset => $location,
598 foreach my $element (@{$value}) {
599 #$branch->[$index] = $element;
600 $branch->STORE( $index, $element );
608 return $self->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
611 sub _get_bucket_value {
613 # Fetch single value given tag and MD5 digested key.
616 my ($tag, $md5) = @_;
617 my $keys = $tag->{content};
622 # Iterate through buckets, looking for a key match
625 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
626 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
627 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
631 # Hit end of list, no match
636 if ( $md5 ne $key ) {
641 # Found match -- seek to offset and read signature
644 seek($fh, $subloc, SEEK_SET);
645 read( $fh, $signature, SIG_SIZE);
648 # If value is a hash or array, return new DeepDB object with correct offset
650 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
651 my $obj = DBM::Deep->new(
653 base_offset => $subloc,
657 if ($self->root->{autobless}) {
659 # Skip over value and plain key to see if object needs
662 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
665 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
666 if ($size) { seek($fh, $size, SEEK_CUR); }
669 read( $fh, $bless_bit, 1);
670 if (ord($bless_bit)) {
672 # Yes, object needs to be re-blessed
675 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
676 if ($size) { read( $fh, $class_name, $size); }
677 if ($class_name) { $obj = bless( $obj, $class_name ); }
685 # Otherwise return actual value
687 elsif ($signature eq SIG_DATA) {
690 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
691 if ($size) { read( $fh, $value, $size); }
696 # Key exists, but content is null
706 # Delete single key/value pair given tag and MD5 digested key.
709 my ($tag, $md5) = @_;
710 my $keys = $tag->{content};
715 # Iterate through buckets, looking for a key match
718 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
719 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
720 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
724 # Hit end of list, no match
729 if ( $md5 ne $key ) {
734 # Matched key -- delete bucket and return
736 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
737 print($fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
738 print($fh chr(0) x $BUCKET_SIZE );
748 # Check existence of single key given tag and MD5 digested key.
751 my ($tag, $md5) = @_;
752 my $keys = $tag->{content};
755 # Iterate through buckets, looking for a key match
758 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
759 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
760 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
764 # Hit end of list, no match
769 if ( $md5 ne $key ) {
774 # Matched key -- return true
782 sub _find_bucket_list {
784 # Locate offset for bucket list, given digested key
790 # Locate offset for bucket list using digest index system
793 my $tag = $self->_load_tag($self->base_offset);
794 if (!$tag) { return; }
796 while ($tag->{signature} ne SIG_BLIST) {
797 $tag = $self->_index_lookup($tag, ord(substr($md5, $ch, 1)));
798 if (!$tag) { return; }
805 sub _traverse_index {
807 # Scan index and recursively step into deeper levels, looking for next key.
809 my ($self, $offset, $ch, $force_return_next) = @_;
810 $force_return_next = undef unless $force_return_next;
812 my $tag = $self->_load_tag( $offset );
816 if ($tag->{signature} ne SIG_BLIST) {
817 my $content = $tag->{content};
819 if ($self->{return_next}) { $start = 0; }
820 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
822 for (my $index = $start; $index < 256; $index++) {
823 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
825 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
826 if (defined($result)) { return $result; }
830 $self->{return_next} = 1;
833 elsif ($tag->{signature} eq SIG_BLIST) {
834 my $keys = $tag->{content};
835 if ($force_return_next) { $self->{return_next} = 1; }
838 # Iterate through buckets, looking for a key match
840 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
841 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
842 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
846 # End of bucket list -- return to outer loop
848 $self->{return_next} = 1;
851 elsif ($key eq $self->{prev_md5}) {
853 # Located previous key -- return next one found
855 $self->{return_next} = 1;
858 elsif ($self->{return_next}) {
860 # Seek to bucket location and skip over signature
862 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
865 # Skip over value to get to plain key
868 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
869 if ($size) { seek($fh, $size, SEEK_CUR); }
872 # Read in plain key and return as scalar
875 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
876 if ($size) { read( $fh, $plain_key, $size); }
882 $self->{return_next} = 1;
883 } # tag is a bucket list
890 # Locate next key, given digested previous one
892 my $self = $_[0]->_get_self;
894 $self->{prev_md5} = $_[1] ? $_[1] : undef;
895 $self->{return_next} = 0;
898 # If the previous key was not specifed, start at the top and
899 # return the first one found.
901 if (!$self->{prev_md5}) {
902 $self->{prev_md5} = chr(0) x $HASH_SIZE;
903 $self->{return_next} = 1;
906 return $self->_traverse_index( $self->base_offset, 0 );
911 # If db locking is set, flock() the db file. If called multiple
912 # times before unlock(), then the same number of unlocks() must
913 # be called before the lock is released.
915 my $self = $_[0]->_get_self;
917 $type = LOCK_EX unless defined $type;
919 if (!defined($self->fh)) { return; }
921 if ($self->root->{locking}) {
922 if (!$self->root->{locked}) { flock($self->fh, $type); }
923 $self->root->{locked}++;
933 # If db locking is set, unlock the db file. See note in lock()
934 # regarding calling lock() multiple times.
936 my $self = $_[0]->_get_self;
938 if (!defined($self->fh)) { return; }
940 if ($self->root->{locking} && $self->root->{locked} > 0) {
941 $self->root->{locked}--;
942 if (!$self->root->{locked}) { flock($self->fh, LOCK_UN); }
950 #XXX These uses of ref() need verified
953 # Copy single level of keys or elements to new DB handle.
954 # Recurse for nested structures
956 my $self = $_[0]->_get_self;
959 if ($self->type eq TYPE_HASH) {
960 my $key = $self->first_key();
962 my $value = $self->get($key);
963 #XXX This doesn't work with autobless
964 if (!ref($value)) { $db_temp->{$key} = $value; }
966 my $type = $value->type;
967 if ($type eq TYPE_HASH) { $db_temp->{$key} = {}; }
968 else { $db_temp->{$key} = []; }
969 $value->_copy_node( $db_temp->{$key} );
971 $key = $self->next_key($key);
975 my $length = $self->length();
976 for (my $index = 0; $index < $length; $index++) {
977 my $value = $self->get($index);
978 if (!ref($value)) { $db_temp->[$index] = $value; }
979 #XXX NO tests for this code
981 my $type = $value->type;
982 if ($type eq TYPE_HASH) { $db_temp->[$index] = {}; }
983 else { $db_temp->[$index] = []; }
984 $value->_copy_node( $db_temp->[$index] );
992 # Recursively export into standard Perl hashes and arrays.
994 my $self = $_[0]->_get_self;
997 if ($self->type eq TYPE_HASH) { $temp = {}; }
998 elsif ($self->type eq TYPE_ARRAY) { $temp = []; }
1001 $self->_copy_node( $temp );
1009 # Recursively import Perl hash/array structure
1011 #XXX This use of ref() seems to be ok
1012 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
1014 my $self = $_[0]->_get_self;
1017 #XXX This use of ref() seems to be ok
1018 if (!ref($struct)) {
1020 # struct is not a reference, so just import based on our type
1024 if ($self->type eq TYPE_HASH) { $struct = {@_}; }
1025 elsif ($self->type eq TYPE_ARRAY) { $struct = [@_]; }
1028 my $r = Scalar::Util::reftype($struct) || '';
1029 if ($r eq "HASH" && $self->type eq TYPE_HASH) {
1030 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
1032 elsif ($r eq "ARRAY" && $self->type eq TYPE_ARRAY) {
1033 $self->push( @$struct );
1036 return $self->_throw_error("Cannot import: type mismatch");
1044 # Rebuild entire database into new file, then move
1045 # it back on top of original.
1047 my $self = $_[0]->_get_self;
1049 #XXX Need to create a new test for this
1050 # if ($self->root->{links} > 1) {
1051 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
1054 my $db_temp = DBM::Deep->new(
1055 file => $self->root->{file} . '.tmp',
1059 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
1063 $self->_copy_node( $db_temp );
1067 # Attempt to copy user, group and permissions over to new file
1069 my @stats = stat($self->fh);
1070 my $perms = $stats[2] & 07777;
1071 my $uid = $stats[4];
1072 my $gid = $stats[5];
1073 chown( $uid, $gid, $self->root->{file} . '.tmp' );
1074 chmod( $perms, $self->root->{file} . '.tmp' );
1076 # q.v. perlport for more information on this variable
1077 if ( $^O eq 'MSWin32' ) {
1079 # Potential race condition when optmizing on Win32 with locking.
1080 # The Windows filesystem requires that the filehandle be closed
1081 # before it is overwritten with rename(). This could be redone
1088 if (!rename $self->root->{file} . '.tmp', $self->root->{file}) {
1089 unlink $self->root->{file} . '.tmp';
1091 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
1103 # Make copy of object and return
1105 my $self = $_[0]->_get_self;
1107 return DBM::Deep->new(
1108 type => $self->type,
1109 base_offset => $self->base_offset,
1115 my %is_legal_filter = map {
1118 store_key store_value
1119 fetch_key fetch_value
1124 # Setup filter function for storing or fetching the key or value
1126 my $self = $_[0]->_get_self;
1127 my $type = lc $_[1];
1128 my $func = $_[2] ? $_[2] : undef;
1130 if ( $is_legal_filter{$type} ) {
1131 $self->root->{"filter_$type"} = $func;
1145 # Get access to the root structure
1147 my $self = $_[0]->_get_self;
1148 return $self->{root};
1153 # Get access to the raw FileHandle
1155 #XXX It will be useful, though, when we split out HASH and ARRAY
1156 my $self = $_[0]->_get_self;
1157 return $self->root->{fh};
1162 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
1164 my $self = $_[0]->_get_self;
1165 return $self->{type};
1170 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
1172 my $self = $_[0]->_get_self;
1173 return $self->{base_offset};
1178 # Get last error string, or undef if no error
1181 #? ( _get_self($_[0])->{root}->{error} or undef )
1182 ? ( $_[0]->_get_self->{root}->{error} or undef )
1192 # Store error string in self
1194 my $self = $_[0]->_get_self;
1195 my $error_text = $_[1];
1197 if ( Scalar::Util::blessed $self ) {
1198 $self->root->{error} = $error_text;
1200 unless ($self->root->{debug}) {
1201 die "DBM::Deep: $error_text\n";
1204 warn "DBM::Deep: $error_text\n";
1208 die "DBM::Deep: $error_text\n";
1216 my $self = $_[0]->_get_self;
1218 undef $self->root->{error};
1223 # Precalculate index, bucket and bucket list sizes
1226 #XXX I don't like this ...
1227 set_pack() unless defined $LONG_SIZE;
1229 $INDEX_SIZE = 256 * $LONG_SIZE;
1230 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
1231 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
1236 # Set pack/unpack modes (see file header for more)
1238 my ($long_s, $long_p, $data_s, $data_p) = @_;
1240 $LONG_SIZE = $long_s ? $long_s : 4;
1241 $LONG_PACK = $long_p ? $long_p : 'N';
1243 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
1244 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
1251 # Set key digest function (default is MD5)
1253 my ($digest_func, $hash_size) = @_;
1255 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
1256 $HASH_SIZE = $hash_size ? $hash_size : 16;
1262 # tie() methods (hashes and arrays)
1267 # Store single hash key/value or array element in database.
1269 my $self = $_[0]->_get_self;
1272 #XXX What is ref() checking here?
1273 #YYY User may be storing a hash, in which case we do not want it run
1274 #YYY through the filtering system
1275 my $value = ($self->root->{filter_store_value} && !ref($_[2]))
1276 ? $self->root->{filter_store_value}->($_[2])
1279 my $md5 = $DIGEST_FUNC->($key);
1282 # Make sure file is open
1284 if (!defined($self->fh) && !$self->_open()) {
1292 # Request exclusive lock for writing
1294 $self->lock( LOCK_EX );
1297 # If locking is enabled, set 'end' parameter again, in case another
1298 # DB instance appended to our file while we were unlocked.
1300 if ($self->root->{locking} || $self->root->{volatile}) {
1301 $self->root->{end} = (stat($fh))[7];
1305 # Locate offset for bucket list using digest index system
1307 my $tag = $self->_load_tag($self->base_offset);
1309 $tag = $self->_create_tag($self->base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1313 while ($tag->{signature} ne SIG_BLIST) {
1314 my $num = ord(substr($md5, $ch, 1));
1315 my $new_tag = $self->_index_lookup($tag, $num);
1317 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1318 seek($fh, $ref_loc, SEEK_SET);
1319 print($fh pack($LONG_PACK, $self->root->{end}) );
1321 $tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1322 $tag->{ref_loc} = $ref_loc;
1327 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1329 $tag->{ref_loc} = $ref_loc;
1336 # Add key/value to bucket list
1338 my $result = $self->_add_bucket( $tag, $md5, $key, $value );
1347 # Fetch single value or element given plain key or array index
1349 my $self = shift->_get_self;
1353 # Make sure file is open
1355 if (!defined($self->fh)) { $self->_open(); }
1357 my $md5 = $DIGEST_FUNC->($key);
1360 # Request shared lock for reading
1362 $self->lock( LOCK_SH );
1364 my $tag = $self->_find_bucket_list( $md5 );
1371 # Get value from bucket list
1373 my $result = $self->_get_bucket_value( $tag, $md5 );
1377 #XXX What is ref() checking here?
1378 return ($result && !ref($result) && $self->root->{filter_fetch_value})
1379 ? $self->root->{filter_fetch_value}->($result)
1385 # Delete single key/value pair or element given plain key or array index
1387 my $self = $_[0]->_get_self;
1390 my $unpacked_key = $key;
1391 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1392 my $md5 = $DIGEST_FUNC->($key);
1395 # Make sure file is open
1397 if (!defined($self->fh)) { $self->_open(); }
1400 # Request exclusive lock for writing
1402 $self->lock( LOCK_EX );
1404 my $tag = $self->_find_bucket_list( $md5 );
1413 my $value = $self->_get_bucket_value( $tag, $md5 );
1414 if ($value && !ref($value) && $self->root->{filter_fetch_value}) {
1415 $value = $self->root->{filter_fetch_value}->($value);
1418 my $result = $self->_delete_bucket( $tag, $md5 );
1421 # If this object is an array and the key deleted was on the end of the stack,
1422 # decrement the length variable.
1424 if ($result && ($self->type eq TYPE_ARRAY) && ($unpacked_key == $self->FETCHSIZE() - 1)) {
1425 $self->STORESIZE( $unpacked_key );
1435 # Check if a single key or element exists given plain key or array index
1437 my $self = $_[0]->_get_self;
1440 my $md5 = $DIGEST_FUNC->($key);
1443 # Make sure file is open
1445 if (!defined($self->fh)) { $self->_open(); }
1448 # Request shared lock for reading
1450 $self->lock( LOCK_SH );
1452 my $tag = $self->_find_bucket_list( $md5 );
1455 # For some reason, the built-in exists() function returns '' for false
1463 # Check if bucket exists and return 1 or ''
1465 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1474 # Clear all keys from hash, or all elements from array.
1476 my $self = $_[0]->_get_self;
1479 # Make sure file is open
1481 if (!defined($self->fh)) { $self->_open(); }
1484 # Request exclusive lock for writing
1486 $self->lock( LOCK_EX );
1490 seek($fh, $self->base_offset, SEEK_SET);
1496 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
1504 # Public method aliases
1506 sub put { (shift)->STORE( @_ ) }
1507 sub store { (shift)->STORE( @_ ) }
1508 sub get { (shift)->FETCH( @_ ) }
1509 sub fetch { (shift)->FETCH( @_ ) }
1510 sub delete { (shift)->DELETE( @_ ) }
1511 sub exists { (shift)->EXISTS( @_ ) }
1512 sub clear { (shift)->CLEAR( @_ ) }
1514 package DBM::Deep::_::Root;
1529 filter_store_key => undef,
1530 filter_store_value => undef,
1531 filter_fetch_key => undef,
1532 filter_fetch_value => undef,
1543 return unless $self;
1545 close $self->{fh} if $self->{fh};
1556 DBM::Deep - A pure perl multi-level hash/array DBM
1561 my $db = DBM::Deep->new( "foo.db" );
1563 $db->{key} = 'value'; # tie() style
1566 $db->put('key', 'value'); # OO style
1567 print $db->get('key');
1569 # true multi-level support
1570 $db->{my_complex} = [
1571 'hello', { perl => 'rules' },
1576 A unique flat-file database module, written in pure perl. True
1577 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1578 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1579 handle millions of keys and unlimited hash levels without significant
1580 slow-down. Written from the ground-up in pure perl -- this is NOT a
1581 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1582 Mac OS X and Windows.
1586 Hopefully you are using CPAN's excellent Perl module, which will download
1587 and install the module for you. If not, get the tarball, and run these
1599 Construction can be done OO-style (which is the recommended way), or using
1600 Perl's tie() function. Both are examined here.
1602 =head2 OO CONSTRUCTION
1604 The recommended way to construct a DBM::Deep object is to use the new()
1605 method, which gets you a blessed, tied hash or array reference.
1607 my $db = DBM::Deep->new( "foo.db" );
1609 This opens a new database handle, mapped to the file "foo.db". If this
1610 file does not exist, it will automatically be created. DB files are
1611 opened in "r+" (read/write) mode, and the type of object returned is a
1612 hash, unless otherwise specified (see L<OPTIONS> below).
1616 You can pass a number of options to the constructor to specify things like
1617 locking, autoflush, etc. This is done by passing an inline hash:
1619 my $db = DBM::Deep->new(
1625 Notice that the filename is now specified I<inside> the hash with
1626 the "file" parameter, as opposed to being the sole argument to the
1627 constructor. This is required if any options are specified.
1628 See L<OPTIONS> below for the complete list.
1632 You can also start with an array instead of a hash. For this, you must
1633 specify the C<type> parameter:
1635 my $db = DBM::Deep->new(
1637 type => DBM::Deep->TYPE_ARRAY
1640 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1641 a new DB file. If you create a DBM::Deep object with an existing file, the
1642 C<type> will be loaded from the file header, and ignored if it is passed
1645 =head2 TIE CONSTRUCTION
1647 Alternatively, you can create a DBM::Deep handle by using Perl's built-in
1648 tie() function. This is not ideal, because you get only a basic, tied hash
1649 (or array) which is not blessed, so you can't call any functions on it.
1652 tie %hash, "DBM::Deep", "foo.db";
1655 tie @array, "DBM::Deep", "bar.db";
1657 As with the OO constructor, you can replace the DB filename parameter with
1658 a hash containing one or more options (see L<OPTIONS> just below for the
1661 tie %hash, "DBM::Deep", {
1669 There are a number of options that can be passed in when constructing your
1670 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1676 Filename of the DB file to link the handle to. You can pass a full absolute
1677 filesystem path, partial path, or a plain filename if the file is in the
1678 current working directory. This is a required parameter.
1682 File open mode (read-only, read-write, etc.) string passed to Perl's FileHandle
1683 module. This is an optional parameter, and defaults to "r+" (read/write).
1684 B<Note:> If the default (r+) mode is selected, the file will also be auto-
1685 created if it doesn't exist.
1689 This parameter specifies what type of object to create, a hash or array. Use
1690 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1691 This only takes effect when beginning a new file. This is an optional
1692 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1696 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1697 function to lock the database in exclusive mode for writes, and shared mode for
1698 reads. Pass any true value to enable. This affects the base DB handle I<and
1699 any child hashes or arrays> that use the same DB file. This is an optional
1700 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1704 Specifies whether autoflush is to be enabled on the underlying FileHandle.
1705 This obviously slows down write operations, but is required if you may have
1706 multiple processes accessing the same DB file (also consider enable I<locking>
1707 or at least I<volatile>). Pass any true value to enable. This is an optional
1708 parameter, and defaults to 0 (disabled).
1712 If I<volatile> mode is enabled, DBM::Deep will stat() the DB file before each
1713 STORE() operation. This is required if an outside force may change the size of
1714 the file between transactions. Locking also implicitly enables volatile. This
1715 is useful if you want to use a different locking system or write your own. Pass
1716 any true value to enable. This is an optional parameter, and defaults to 0
1721 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1722 restore them when fetched. This is an B<experimental> feature, and does have
1723 side-effects. Basically, when hashes are re-blessed into their original
1724 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1725 able to call any DBM::Deep methods on them. You have been warned.
1726 This is an optional parameter, and defaults to 0 (disabled).
1730 See L<FILTERS> below.
1734 Setting I<debug> mode will make all errors non-fatal, dump them out to
1735 STDERR, and continue on. This is for debugging purposes only, and probably
1736 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1740 Instead of passing a file path, you can instead pass a handle to an pre-opened
1741 filehandle. Note: Beware of using the magick *DATA handle, as this actually
1742 contains your entire Perl script, as well as the data following the __DATA__
1743 marker. This will not work, because DBM::Deep uses absolute seek()s into the
1744 file. Instead, consider reading *DATA into an IO::Scalar handle, then passing
1749 =head1 TIE INTERFACE
1751 With DBM::Deep you can access your databases using Perl's standard hash/array
1752 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can treat
1753 them as such. DBM::Deep will intercept all reads/writes and direct them to the right
1754 place -- the DB file. This has nothing to do with the L<TIE CONSTRUCTION>
1755 section above. This simply tells you how to use DBM::Deep using regular hashes
1756 and arrays, rather than calling functions like C<get()> and C<put()> (although those
1757 work too). It is entirely up to you how to want to access your databases.
1761 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1762 or even nested hashes (or arrays) using standard Perl syntax:
1764 my $db = DBM::Deep->new( "foo.db" );
1766 $db->{mykey} = "myvalue";
1768 $db->{myhash}->{subkey} = "subvalue";
1770 print $db->{myhash}->{subkey} . "\n";
1772 You can even step through hash keys using the normal Perl C<keys()> function:
1774 foreach my $key (keys %$db) {
1775 print "$key: " . $db->{$key} . "\n";
1778 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1779 pushes them onto an array, all before the loop even begins. If you have an
1780 extra large hash, this may exhaust Perl's memory. Instead, consider using
1781 Perl's C<each()> function, which pulls keys/values one at a time, using very
1784 while (my ($key, $value) = each %$db) {
1785 print "$key: $value\n";
1788 Please note that when using C<each()>, you should always pass a direct
1789 hash reference, not a lookup. Meaning, you should B<never> do this:
1792 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1794 This causes an infinite loop, because for each iteration, Perl is calling
1795 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1796 it effectively keeps returning the first key over and over again. Instead,
1797 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1801 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1802 reference. This includes inserting, removing and manipulating elements,
1803 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1804 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1805 or simply be a nested array reference inside a hash. Example:
1807 my $db = DBM::Deep->new(
1808 file => "foo-array.db",
1809 type => DBM::Deep->TYPE_ARRAY
1813 push @$db, "bar", "baz";
1814 unshift @$db, "bah";
1816 my $last_elem = pop @$db; # baz
1817 my $first_elem = shift @$db; # bah
1818 my $second_elem = $db->[1]; # bar
1820 my $num_elements = scalar @$db;
1824 In addition to the I<tie()> interface, you can also use a standard OO interface
1825 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1826 array) has its own methods, but both types share the following common methods:
1827 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1833 Stores a new hash key/value pair, or sets an array element value. Takes two
1834 arguments, the hash key or array index, and the new value. The value can be
1835 a scalar, hash ref or array ref. Returns true on success, false on failure.
1837 $db->put("foo", "bar"); # for hashes
1838 $db->put(1, "bar"); # for arrays
1842 Fetches the value of a hash key or array element. Takes one argument: the hash
1843 key or array index. Returns a scalar, hash ref or array ref, depending on the
1846 my $value = $db->get("foo"); # for hashes
1847 my $value = $db->get(1); # for arrays
1851 Checks if a hash key or array index exists. Takes one argument: the hash key
1852 or array index. Returns true if it exists, false if not.
1854 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1855 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1859 Deletes one hash key/value pair or array element. Takes one argument: the hash
1860 key or array index. Returns true on success, false if not found. For arrays,
1861 the remaining elements located after the deleted element are NOT moved over.
1862 The deleted element is essentially just undefined, which is exactly how Perl's
1863 internal arrays work. Please note that the space occupied by the deleted
1864 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1865 below for details and workarounds.
1867 $db->delete("foo"); # for hashes
1868 $db->delete(1); # for arrays
1872 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1873 value. Please note that the space occupied by the deleted keys/values or
1874 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1875 details and workarounds.
1877 $db->clear(); # hashes or arrays
1883 For hashes, DBM::Deep supports all the common methods described above, and the
1884 following additional methods: C<first_key()> and C<next_key()>.
1890 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1891 fetched in an undefined order (which appears random). Takes no arguments,
1892 returns the key as a scalar value.
1894 my $key = $db->first_key();
1898 Returns the "next" key in the hash, given the previous one as the sole argument.
1899 Returns undef if there are no more keys to be fetched.
1901 $key = $db->next_key($key);
1905 Here are some examples of using hashes:
1907 my $db = DBM::Deep->new( "foo.db" );
1909 $db->put("foo", "bar");
1910 print "foo: " . $db->get("foo") . "\n";
1912 $db->put("baz", {}); # new child hash ref
1913 $db->get("baz")->put("buz", "biz");
1914 print "buz: " . $db->get("baz")->get("buz") . "\n";
1916 my $key = $db->first_key();
1918 print "$key: " . $db->get($key) . "\n";
1919 $key = $db->next_key($key);
1922 if ($db->exists("foo")) { $db->delete("foo"); }
1926 For arrays, DBM::Deep supports all the common methods described above, and the
1927 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1928 C<unshift()> and C<splice()>.
1934 Returns the number of elements in the array. Takes no arguments.
1936 my $len = $db->length();
1940 Adds one or more elements onto the end of the array. Accepts scalars, hash
1941 refs or array refs. No return value.
1943 $db->push("foo", "bar", {});
1947 Fetches the last element in the array, and deletes it. Takes no arguments.
1948 Returns undef if array is empty. Returns the element value.
1950 my $elem = $db->pop();
1954 Fetches the first element in the array, deletes it, then shifts all the
1955 remaining elements over to take up the space. Returns the element value. This
1956 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1959 my $elem = $db->shift();
1963 Inserts one or more elements onto the beginning of the array, shifting all
1964 existing elements over to make room. Accepts scalars, hash refs or array refs.
1965 No return value. This method is not recommended with large arrays -- see
1966 <LARGE ARRAYS> below for details.
1968 $db->unshift("foo", "bar", {});
1972 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1973 -f splice> for usage -- it is too complicated to document here. This method is
1974 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1978 Here are some examples of using arrays:
1980 my $db = DBM::Deep->new(
1982 type => DBM::Deep->TYPE_ARRAY
1985 $db->push("bar", "baz");
1986 $db->unshift("foo");
1989 my $len = $db->length();
1990 print "length: $len\n"; # 4
1992 for (my $k=0; $k<$len; $k++) {
1993 print "$k: " . $db->get($k) . "\n";
1996 $db->splice(1, 2, "biz", "baf");
1998 while (my $elem = shift @$db) {
1999 print "shifted: $elem\n";
2004 Enable automatic file locking by passing a true value to the C<locking>
2005 parameter when constructing your DBM::Deep object (see L<SETUP> above).
2007 my $db = DBM::Deep->new(
2012 This causes DBM::Deep to C<flock()> the underlying FileHandle object with exclusive
2013 mode for writes, and shared mode for reads. This is required if you have
2014 multiple processes accessing the same database file, to avoid file corruption.
2015 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
2016 NFS> below for more.
2018 =head2 EXPLICIT LOCKING
2020 You can explicitly lock a database, so it remains locked for multiple
2021 transactions. This is done by calling the C<lock()> method, and passing an
2022 optional lock mode argument (defaults to exclusive mode). This is particularly
2023 useful for things like counters, where the current value needs to be fetched,
2024 then incremented, then stored again.
2027 my $counter = $db->get("counter");
2029 $db->put("counter", $counter);
2038 You can pass C<lock()> an optional argument, which specifies which mode to use
2039 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2040 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2041 same as the constants defined in Perl's C<Fcntl> module.
2043 $db->lock( DBM::Deep->LOCK_SH );
2047 If you want to implement your own file locking scheme, be sure to create your
2048 DBM::Deep objects setting the C<volatile> option to true. This hints to DBM::Deep
2049 that the DB file may change between transactions. See L<LOW-LEVEL ACCESS>
2052 =head1 IMPORTING/EXPORTING
2054 You can import existing complex structures by calling the C<import()> method,
2055 and export an entire database into an in-memory structure using the C<export()>
2056 method. Both are examined here.
2060 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2061 walking the structure and adding keys/elements to the database as you go,
2062 simply pass a reference to the C<import()> method. This recursively adds
2063 everything to an existing DBM::Deep object for you. Here is an example:
2068 array1 => [ "elem0", "elem1", "elem2" ],
2070 subkey1 => "subvalue1",
2071 subkey2 => "subvalue2"
2075 my $db = DBM::Deep->new( "foo.db" );
2076 $db->import( $struct );
2078 print $db->{key1} . "\n"; # prints "value1"
2080 This recursively imports the entire C<$struct> object into C<$db>, including
2081 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2082 keys are merged with the existing ones, replacing if they already exist.
2083 The C<import()> method can be called on any database level (not just the base
2084 level), and works with both hash and array DB types.
2088 B<Note:> Make sure your existing structure has no circular references in it.
2089 These will cause an infinite loop when importing.
2093 Calling the C<export()> method on an existing DBM::Deep object will return
2094 a reference to a new in-memory copy of the database. The export is done
2095 recursively, so all nested hashes/arrays are all exported to standard Perl
2096 objects. Here is an example:
2098 my $db = DBM::Deep->new( "foo.db" );
2100 $db->{key1} = "value1";
2101 $db->{key2} = "value2";
2103 $db->{hash1}->{subkey1} = "subvalue1";
2104 $db->{hash1}->{subkey2} = "subvalue2";
2106 my $struct = $db->export();
2108 print $struct->{key1} . "\n"; # prints "value1"
2110 This makes a complete copy of the database in memory, and returns a reference
2111 to it. The C<export()> method can be called on any database level (not just
2112 the base level), and works with both hash and array DB types. Be careful of
2113 large databases -- you can store a lot more data in a DBM::Deep object than an
2114 in-memory Perl structure.
2118 B<Note:> Make sure your database has no circular references in it.
2119 These will cause an infinite loop when exporting.
2123 DBM::Deep has a number of hooks where you can specify your own Perl function
2124 to perform filtering on incoming or outgoing data. This is a perfect
2125 way to extend the engine, and implement things like real-time compression or
2126 encryption. Filtering applies to the base DB level, and all child hashes /
2127 arrays. Filter hooks can be specified when your DBM::Deep object is first
2128 constructed, or by calling the C<set_filter()> method at any time. There are
2129 four available filter hooks, described below:
2133 =item * filter_store_key
2135 This filter is called whenever a hash key is stored. It
2136 is passed the incoming key, and expected to return a transformed key.
2138 =item * filter_store_value
2140 This filter is called whenever a hash key or array element is stored. It
2141 is passed the incoming value, and expected to return a transformed value.
2143 =item * filter_fetch_key
2145 This filter is called whenever a hash key is fetched (i.e. via
2146 C<first_key()> or C<next_key()>). It is passed the transformed key,
2147 and expected to return the plain key.
2149 =item * filter_fetch_value
2151 This filter is called whenever a hash key or array element is fetched.
2152 It is passed the transformed value, and expected to return the plain value.
2156 Here are the two ways to setup a filter hook:
2158 my $db = DBM::Deep->new(
2160 filter_store_value => \&my_filter_store,
2161 filter_fetch_value => \&my_filter_fetch
2166 $db->set_filter( "filter_store_value", \&my_filter_store );
2167 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2169 Your filter function will be called only when dealing with SCALAR keys or
2170 values. When nested hashes and arrays are being stored/fetched, filtering
2171 is bypassed. Filters are called as static functions, passed a single SCALAR
2172 argument, and expected to return a single SCALAR value. If you want to
2173 remove a filter, set the function reference to C<undef>:
2175 $db->set_filter( "filter_store_value", undef );
2177 =head2 REAL-TIME ENCRYPTION EXAMPLE
2179 Here is a working example that uses the I<Crypt::Blowfish> module to
2180 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2181 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2182 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2185 use Crypt::Blowfish;
2188 my $cipher = Crypt::CBC->new({
2189 'key' => 'my secret key',
2190 'cipher' => 'Blowfish',
2192 'regenerate_key' => 0,
2193 'padding' => 'space',
2197 my $db = DBM::Deep->new(
2198 file => "foo-encrypt.db",
2199 filter_store_key => \&my_encrypt,
2200 filter_store_value => \&my_encrypt,
2201 filter_fetch_key => \&my_decrypt,
2202 filter_fetch_value => \&my_decrypt,
2205 $db->{key1} = "value1";
2206 $db->{key2} = "value2";
2207 print "key1: " . $db->{key1} . "\n";
2208 print "key2: " . $db->{key2} . "\n";
2214 return $cipher->encrypt( $_[0] );
2217 return $cipher->decrypt( $_[0] );
2220 =head2 REAL-TIME COMPRESSION EXAMPLE
2222 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2223 compression / decompression of keys & values with DBM::Deep Filters.
2224 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2225 more on I<Compress::Zlib>.
2230 my $db = DBM::Deep->new(
2231 file => "foo-compress.db",
2232 filter_store_key => \&my_compress,
2233 filter_store_value => \&my_compress,
2234 filter_fetch_key => \&my_decompress,
2235 filter_fetch_value => \&my_decompress,
2238 $db->{key1} = "value1";
2239 $db->{key2} = "value2";
2240 print "key1: " . $db->{key1} . "\n";
2241 print "key2: " . $db->{key2} . "\n";
2247 return Compress::Zlib::memGzip( $_[0] ) ;
2250 return Compress::Zlib::memGunzip( $_[0] ) ;
2253 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2254 actually numerical index numbers, and are not filtered.
2256 =head1 ERROR HANDLING
2258 Most DBM::Deep methods return a true value for success, and call die() on
2259 failure. You can wrap calls in an eval block to catch the die. Also, the
2260 actual error message is stored in an internal scalar, which can be fetched by
2261 calling the C<error()> method.
2263 my $db = DBM::Deep->new( "foo.db" ); # create hash
2264 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2266 print $db->error(); # prints error message
2268 You can then call C<clear_error()> to clear the current error state.
2272 If you set the C<debug> option to true when creating your DBM::Deep object,
2273 all errors are considered NON-FATAL, and dumped to STDERR. This is only
2274 for debugging purposes.
2276 =head1 LARGEFILE SUPPORT
2278 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2279 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2280 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2281 by calling the static C<set_pack()> method before you do anything else.
2283 DBM::Deep::set_pack(8, 'Q');
2285 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2286 instead of 32-bit longs. After setting these values your DB files have a
2287 theoretical maximum size of 16 XB (exabytes).
2291 B<Note:> Changing these values will B<NOT> work for existing database files.
2292 Only change this for new files, and make sure it stays set consistently
2293 throughout the file's life. If you do set these values, you can no longer
2294 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2295 back to 32-bit mode.
2299 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2300 only a 32-bit Perl. However, I have received user reports that this does
2303 =head1 LOW-LEVEL ACCESS
2305 If you require low-level access to the underlying FileHandle that DBM::Deep uses,
2306 you can call the C<fh()> method, which returns the handle:
2310 This method can be called on the root level of the datbase, or any child
2311 hashes or arrays. All levels share a I<root> structure, which contains things
2312 like the FileHandle, a reference counter, and all your options you specified
2313 when you created the object. You can get access to this root structure by
2314 calling the C<root()> method.
2316 my $root = $db->root();
2318 This is useful for changing options after the object has already been created,
2319 such as enabling/disabling locking, volatile or debug modes. You can also
2320 store your own temporary user data in this structure (be wary of name
2321 collision), which is then accessible from any child hash or array.
2323 =head1 CUSTOM DIGEST ALGORITHM
2325 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2326 keys. However you can override this, and use another algorithm (such as SHA-256)
2327 or even write your own. But please note that DBM::Deep currently expects zero
2328 collisions, so your algorithm has to be I<perfect>, so to speak.
2329 Collision detection may be introduced in a later version.
2333 You can specify a custom digest algorithm by calling the static C<set_digest()>
2334 function, passing a reference to a subroutine, and the length of the algorithm's
2335 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2336 objects. Here is a working example that uses a 256-bit hash from the
2337 I<Digest::SHA256> module. Please see
2338 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2343 my $context = Digest::SHA256::new(256);
2345 DBM::Deep::set_digest( \&my_digest, 32 );
2347 my $db = DBM::Deep->new( "foo-sha.db" );
2349 $db->{key1} = "value1";
2350 $db->{key2} = "value2";
2351 print "key1: " . $db->{key1} . "\n";
2352 print "key2: " . $db->{key2} . "\n";
2358 return substr( $context->hash($_[0]), 0, 32 );
2361 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2362 of bytes you specify in the C<set_digest()> function (in this case 32).
2364 =head1 CIRCULAR REFERENCES
2366 DBM::Deep has B<experimental> support for circular references. Meaning you
2367 can have a nested hash key or array element that points to a parent object.
2368 This relationship is stored in the DB file, and is preserved between sessions.
2371 my $db = DBM::Deep->new( "foo.db" );
2374 $db->{circle} = $db; # ref to self
2376 print $db->{foo} . "\n"; # prints "foo"
2377 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2379 One catch is, passing the object to a function that recursively walks the
2380 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2381 C<export()> methods) will result in an infinite loop. The other catch is,
2382 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2383 or C<next_key()> methods), you will get the I<target object's key>, not the
2384 ref's key. This gets even more interesting with the above example, where
2385 the I<circle> key points to the base DB object, which technically doesn't
2386 have a key. So I made DBM::Deep return "[base]" as the key name in that
2389 =head1 CAVEATS / ISSUES / BUGS
2391 This section describes all the known issues with DBM::Deep. It you have found
2392 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2394 =head2 UNUSED SPACE RECOVERY
2396 One major caveat with DBM::Deep is that space occupied by existing keys and
2397 values is not recovered when they are deleted. Meaning if you keep deleting
2398 and adding new keys, your file will continuously grow. I am working on this,
2399 but in the meantime you can call the built-in C<optimize()> method from time to
2400 time (perhaps in a crontab or something) to recover all your unused space.
2402 $db->optimize(); # returns true on success
2404 This rebuilds the ENTIRE database into a new file, then moves it on top of
2405 the original. The new file will have no unused space, thus it will take up as
2406 little disk space as possible. Please note that this operation can take
2407 a long time for large files, and you need enough disk space to temporarily hold
2408 2 copies of your DB file. The temporary file is created in the same directory
2409 as the original, named with a ".tmp" extension, and is deleted when the
2410 operation completes. Oh, and if locking is enabled, the DB is automatically
2411 locked for the entire duration of the copy.
2415 B<WARNING:> Only call optimize() on the top-level node of the database, and
2416 make sure there are no child references lying around. DBM::Deep keeps a reference
2417 counter, and if it is greater than 1, optimize() will abort and return undef.
2419 =head2 AUTOVIVIFICATION
2421 Unfortunately, autovivification doesn't work with tied hashes. This appears to
2422 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
2423 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
2424 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
2425 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
2428 $db->{foo}->{bar} = "hello";
2430 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
2431 being an empty hash. Try this instead, which works fine:
2433 $db->{foo} = { bar => "hello" };
2435 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
2436 the execution path, and Perl indeed passes an empty hash to the STORE() method.
2437 Probably a bug in Perl.
2439 =head2 FILE CORRUPTION
2441 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2442 for a 32-bit signature when opened, but other corruption in files can cause
2443 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2444 stuck in an infinite loop depending on the level of corruption. File write
2445 operations are not checked for failure (for speed), so if you happen to run
2446 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2447 be addressed in a later version of DBM::Deep.
2451 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2452 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2453 about setting up your NFS server with a locking daemon, then using lockf() to
2454 lock your files, but your milage may vary there as well. From what I
2455 understand, there is no real way to do it. However, if you need access to the
2456 underlying FileHandle in DBM::Deep for using some other kind of locking scheme like
2457 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2459 =head2 COPYING OBJECTS
2461 Beware of copying tied objects in Perl. Very strange things can happen.
2462 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2463 returns a new, blessed, tied hash or array to the same level in the DB.
2465 my $copy = $db->clone();
2469 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2470 These functions cause every element in the array to move, which can be murder
2471 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2472 a different location. This may be addressed in a later version.
2476 This section discusses DBM::Deep's speed and memory usage.
2480 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2481 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2482 multi-level hash/array support, and cross-platform FTPable files. Even so,
2483 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2484 with huge databases. Here is some test data:
2486 Adding 1,000,000 keys to new DB file...
2488 At 100 keys, avg. speed is 2,703 keys/sec
2489 At 200 keys, avg. speed is 2,642 keys/sec
2490 At 300 keys, avg. speed is 2,598 keys/sec
2491 At 400 keys, avg. speed is 2,578 keys/sec
2492 At 500 keys, avg. speed is 2,722 keys/sec
2493 At 600 keys, avg. speed is 2,628 keys/sec
2494 At 700 keys, avg. speed is 2,700 keys/sec
2495 At 800 keys, avg. speed is 2,607 keys/sec
2496 At 900 keys, avg. speed is 2,190 keys/sec
2497 At 1,000 keys, avg. speed is 2,570 keys/sec
2498 At 2,000 keys, avg. speed is 2,417 keys/sec
2499 At 3,000 keys, avg. speed is 1,982 keys/sec
2500 At 4,000 keys, avg. speed is 1,568 keys/sec
2501 At 5,000 keys, avg. speed is 1,533 keys/sec
2502 At 6,000 keys, avg. speed is 1,787 keys/sec
2503 At 7,000 keys, avg. speed is 1,977 keys/sec
2504 At 8,000 keys, avg. speed is 2,028 keys/sec
2505 At 9,000 keys, avg. speed is 2,077 keys/sec
2506 At 10,000 keys, avg. speed is 2,031 keys/sec
2507 At 20,000 keys, avg. speed is 1,970 keys/sec
2508 At 30,000 keys, avg. speed is 2,050 keys/sec
2509 At 40,000 keys, avg. speed is 2,073 keys/sec
2510 At 50,000 keys, avg. speed is 1,973 keys/sec
2511 At 60,000 keys, avg. speed is 1,914 keys/sec
2512 At 70,000 keys, avg. speed is 2,091 keys/sec
2513 At 80,000 keys, avg. speed is 2,103 keys/sec
2514 At 90,000 keys, avg. speed is 1,886 keys/sec
2515 At 100,000 keys, avg. speed is 1,970 keys/sec
2516 At 200,000 keys, avg. speed is 2,053 keys/sec
2517 At 300,000 keys, avg. speed is 1,697 keys/sec
2518 At 400,000 keys, avg. speed is 1,838 keys/sec
2519 At 500,000 keys, avg. speed is 1,941 keys/sec
2520 At 600,000 keys, avg. speed is 1,930 keys/sec
2521 At 700,000 keys, avg. speed is 1,735 keys/sec
2522 At 800,000 keys, avg. speed is 1,795 keys/sec
2523 At 900,000 keys, avg. speed is 1,221 keys/sec
2524 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2526 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2527 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2528 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2529 Run time was 12 min 3 sec.
2533 One of the great things about DBM::Deep is that it uses very little memory.
2534 Even with huge databases (1,000,000+ keys) you will not see much increased
2535 memory on your process. DBM::Deep relies solely on the filesystem for storing
2536 and fetching data. Here is output from I</usr/bin/top> before even opening a
2539 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2540 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2542 Basically the process is taking 2,716K of memory. And here is the same
2543 process after storing and fetching 1,000,000 keys:
2545 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2546 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2548 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2549 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2551 =head1 DB FILE FORMAT
2553 In case you were interested in the underlying DB file format, it is documented
2554 here in this section. You don't need to know this to use the module, it's just
2555 included for reference.
2559 DBM::Deep files always start with a 32-bit signature to identify the file type.
2560 This is at offset 0. The signature is "DPDB" in network byte order. This is
2561 checked when the file is opened.
2565 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2566 has a standard header containing the type of data, the length of data, and then
2567 the data itself. The type is a single character (1 byte), the length is a
2568 32-bit unsigned long in network byte order, and the data is, well, the data.
2569 Here is how it unfolds:
2573 Immediately after the 32-bit file signature is the I<Master Index> record.
2574 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2575 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2576 depending on how the DBM::Deep object was constructed.
2580 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2581 number). The first 8-bit char of the MD5 signature is the offset into the
2582 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2583 index element is a file offset of the next tag for the key/element in question,
2584 which is usually a I<Bucket List> tag (see below).
2588 The next tag I<could> be another index, depending on how many keys/elements
2589 exist. See L<RE-INDEXING> below for details.
2593 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2594 file offsets to where the actual data is stored. It starts with a standard
2595 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2596 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2597 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2598 When the list fills up, a I<Re-Index> operation is performed (See
2599 L<RE-INDEXING> below).
2603 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2604 index/value pair (in array mode). It starts with a standard tag header with
2605 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2606 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2607 header. The size reported in the tag header is only for the value, but then,
2608 just after the value is another size (32-bit unsigned long) and then the plain
2609 key itself. Since the value is likely to be fetched more often than the plain
2610 key, I figured it would be I<slightly> faster to store the value first.
2614 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2615 record for the nested structure, where the process begins all over again.
2619 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2620 exhausted. Then, when another key/element comes in, the list is converted to a
2621 new index record. However, this index will look at the next char in the MD5
2622 hash, and arrange new Bucket List pointers accordingly. This process is called
2623 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2624 17 (16 + new one) keys/elements are removed from the old Bucket List and
2625 inserted into the new index. Several new Bucket Lists are created in the
2626 process, as a new MD5 char from the key is being examined (it is unlikely that
2627 the keys will all share the same next char of their MD5s).
2631 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2632 when the Bucket Lists will turn into indexes, but the first round tends to
2633 happen right around 4,000 keys. You will see a I<slight> decrease in
2634 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2635 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2636 right around 900,000 keys. This process can continue nearly indefinitely --
2637 right up until the point the I<MD5> signatures start colliding with each other,
2638 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2639 getting struck by lightning while you are walking to cash in your tickets.
2640 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2641 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2642 this is 340 unodecillion, but don't quote me).
2646 When a new key/element is stored, the key (or index number) is first ran through
2647 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2648 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2649 for the first char of the signature (in this case I<b>). If it does not exist,
2650 a new I<Bucket List> is created for our key (and the next 15 future keys that
2651 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2652 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2653 this point, unless we are replacing an existing I<Bucket>), where the actual
2654 data will be stored.
2658 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2659 (or index number), then walking along the indexes. If there are enough
2660 keys/elements in this DB level, there might be nested indexes, each linked to
2661 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2662 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2663 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2664 plain key are stored.
2668 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2669 methods. In this process the indexes are walked systematically, and each key
2670 fetched in increasing MD5 order (which is why it appears random). Once the
2671 I<Bucket> is found, the value is skipped the plain key returned instead.
2672 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2673 alphabetically sorted. This only happens on an index-level -- as soon as the
2674 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2675 so it's pretty much undefined how the keys will come out -- just like Perl's
2678 =head1 CODE COVERAGE
2680 I use B<Devel::Cover> to test the code coverage of my tests, below is the B<Devel::Cover> report on this
2681 module's test suite.
2683 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2684 File stmt bran cond sub pod time total
2685 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2686 blib/lib/DBM/Deep.pm 93.9 82.4 74.7 97.9 10.5 85.7 88.0
2687 blib/lib/DBM/Deep/Array.pm 97.8 84.6 50.0 100.0 n/a 9.0 94.6
2688 blib/lib/DBM/Deep/Hash.pm 93.9 87.5 100.0 100.0 n/a 5.3 93.4
2689 Total 94.4 82.9 75.8 98.5 10.5 100.0 89.0
2690 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2694 Joseph Huckaby, L<jhuckaby@cpan.org>
2696 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2700 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2701 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2705 Copyright (c) 2002-2005 Joseph Huckaby. All Rights Reserved.
2706 This is free software, you may use it and distribute it under the
2707 same terms as Perl itself.