7 # Multi-level database module for storing hash trees, arrays and simple
8 # key/value pairs into FTP-able, cross-platform binary database files.
10 # Type `perldoc DBM::Deep` for complete documentation.
14 # tie %db, 'DBM::Deep', 'my_database.db'; # standard tie() method
16 # my $db = new DBM::Deep( 'my_database.db' ); # preferred OO method
18 # $db->{my_scalar} = 'hello world';
19 # $db->{my_hash} = { larry => 'genius', hashes => 'fast' };
20 # $db->{my_array} = [ 1, 2, 3, time() ];
21 # $db->{my_complex} = [ 'hello', { perl => 'rules' }, 42, 99 ];
22 # push @{$db->{my_array}}, 'another value';
23 # my @key_list = keys %{$db->{my_hash}};
24 # print "This module " . $db->{my_complex}->[1]->{perl} . "!\n";
27 # (c) 2002-2005 Joseph Huckaby. All Rights Reserved.
28 # This program is free software; you can redistribute it and/or
29 # modify it under the same terms as Perl itself.
34 use Fcntl qw( :DEFAULT :flock :seek );
38 use vars qw( $VERSION );
42 # Set to 4 and 'N' for 32-bit offset tags (default). Theoretical limit of 4 GB per file.
43 # (Perl must be compiled with largefile support for files > 2 GB)
45 # Set to 8 and 'Q' for 64-bit offsets. Theoretical limit of 16 XB per file.
46 # (Perl must be compiled with largefile and 64-bit long support)
52 # Set to 4 and 'N' for 32-bit data length prefixes. Limit of 4 GB for each key/value.
53 # Upgrading this is possible (see above) but probably not necessary. If you need
54 # more than 4 GB for a single key or value, this module is really not for you :-)
56 #my $DATA_LENGTH_SIZE = 4;
57 #my $DATA_LENGTH_PACK = 'N';
58 our ($LONG_SIZE, $LONG_PACK, $DATA_LENGTH_SIZE, $DATA_LENGTH_PACK);
61 # Maximum number of buckets per list before another level of indexing is done.
62 # Increase this value for slightly greater speed, but larger database files.
63 # DO NOT decrease this value below 16, due to risk of recursive reindex overrun.
68 # Better not adjust anything below here, unless you're me :-)
72 # Setup digest function for keys
74 our ($DIGEST_FUNC, $HASH_SIZE);
75 #my $DIGEST_FUNC = \&Digest::MD5::md5;
78 # Precalculate index and bucket sizes based on values above.
81 my ($INDEX_SIZE, $BUCKET_SIZE, $BUCKET_LIST_SIZE);
88 # Setup file and tag signatures. These should never change.
90 sub SIG_FILE () { 'DPDB' }
91 sub SIG_HASH () { 'H' }
92 sub SIG_ARRAY () { 'A' }
93 sub SIG_SCALAR () { 'S' }
94 sub SIG_NULL () { 'N' }
95 sub SIG_DATA () { 'D' }
96 sub SIG_INDEX () { 'I' }
97 sub SIG_BLIST () { 'B' }
101 # Setup constants for users to pass to new()
103 sub TYPE_HASH () { return SIG_HASH; }
104 sub TYPE_ARRAY () { return SIG_ARRAY; }
105 sub TYPE_SCALAR () { return SIG_SCALAR; }
111 if (scalar(@_) > 1) {
113 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
117 elsif ( my $type = Scalar::Util::reftype($_[0]) ) {
118 if ( $type ne 'HASH' ) {
119 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
124 $args = { file => shift };
132 # Class constructor method for Perl OO interface.
133 # Calls tie() and returns blessed reference to tied hash or array,
134 # providing a hybrid OO/tie interface.
137 my $args = $class->_get_args( @_ );
140 # Check if we want a tied hash or array.
143 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
144 $class = 'DBM::Deep::Array';
145 require DBM::Deep::Array;
146 tie @$self, $class, %$args;
149 $class = 'DBM::Deep::Hash';
150 require DBM::Deep::Hash;
151 tie %$self, $class, %$args;
154 return bless $self, $class;
159 # Setup $self and bless into this class.
164 # These are the defaults to be optionally overridden below
167 base_offset => length(SIG_FILE),
170 foreach my $param ( keys %$self ) {
171 next unless exists $args->{$param};
172 $self->{$param} = delete $args->{$param}
175 $self->{root} = exists $args->{root}
177 : DBM::Deep::_::Root->new( $args );
179 if (!defined($self->fh)) { $self->_open(); }
186 require DBM::Deep::Hash;
187 return DBM::Deep::Hash->TIEHASH( @_ );
192 require DBM::Deep::Array;
193 return DBM::Deep::Array->TIEARRAY( @_ );
196 #XXX Unneeded now ...
202 # Open a FileHandle to the database, create if nonexistent.
203 # Make sure file signature matches DeepDB spec.
205 my $self = $_[0]->_get_self;
207 if (defined($self->fh)) { $self->_close(); }
210 # Theoretically, adding O_BINARY should remove the need for the binmode
211 # Of course, testing it is going to be ... interesting.
212 my $flags = O_RDWR | O_CREAT | O_BINARY;
214 #XXX Can the mode be anything but r+, w+, or a+??
215 #XXX ie, it has to be in read-write mode
216 #XXX So, should we verify that the mode is legitimate?
218 #XXX Maybe the mode thingy should just go away. There's no good
219 #XXX reason for it ...
220 if ( $self->root->{mode} eq 'w+' ) {
225 sysopen( $fh, $self->root->{file}, $flags )
227 $self->root->{fh} = $fh;
228 }; if ($@ ) { $self->_throw_error( "Received error: $@\n" ); }
229 if (! defined($self->fh)) {
230 return $self->_throw_error("Cannot sysopen file: " . $self->root->{file} . ": $!");
235 #XXX Can we remove this by using the right sysopen() flags?
236 # Maybe ... q.v. above
237 binmode $fh; # for win32
239 if ($self->root->{autoflush}) {
240 my $old = select $fh;
246 seek($fh, 0, SEEK_SET);
249 my $bytes_read = read( $fh, $signature, length(SIG_FILE));
252 # File is empty -- write signature and master index
255 seek($fh, 0, SEEK_SET);
257 $self->root->{end} = length(SIG_FILE);
258 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
260 my $plain_key = "[base]";
261 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
262 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
264 # Flush the filehandle
265 my $old_fh = select $fh;
275 # Check signature was valid
277 unless ($signature eq SIG_FILE) {
279 return $self->_throw_error("Signature not found -- file is not a Deep DB");
282 my @stats = stat($fh);
283 $self->root->{inode} = $stats[1];
284 $self->root->{end} = $stats[7];
287 # Get our type from master index signature
289 my $tag = $self->_load_tag($self->base_offset);
291 #XXX We probably also want to store the hash algorithm name and not assume anything
292 #XXX The cool thing would be to allow a different hashing algorithm at every level
295 return $self->_throw_error("Corrupted file, no master index record");
297 if ($self->{type} ne $tag->{signature}) {
298 return $self->_throw_error("File type mismatch");
306 # Close database FileHandle
308 my $self = $_[0]->_get_self;
309 close $self->root->{fh};
310 $self->root->{fh} = undef;
315 # Given offset, signature and content, create tag and write to disk
317 my ($self, $offset, $sig, $content) = @_;
318 my $size = length($content);
322 seek($fh, $offset, SEEK_SET);
323 print($fh $sig . pack($DATA_LENGTH_PACK, $size) . $content );
325 if ($offset == $self->root->{end}) {
326 $self->root->{end} += SIG_SIZE + $DATA_LENGTH_SIZE + $size;
332 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
339 # Given offset, load single tag and return signature, size and data
346 seek($fh, $offset, SEEK_SET);
347 if (eof $fh) { return undef; }
350 read( $fh, $sig, SIG_SIZE);
353 read( $fh, $size, $DATA_LENGTH_SIZE);
354 $size = unpack($DATA_LENGTH_PACK, $size);
357 read( $fh, $buffer, $size);
362 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
369 # Given index tag, lookup single entry in index and return .
372 my ($tag, $index) = @_;
374 my $location = unpack($LONG_PACK, substr($tag->{content}, $index * $LONG_SIZE, $LONG_SIZE) );
375 if (!$location) { return; }
377 return $self->_load_tag( $location );
382 # Adds one key/value pair to bucket list, given offset, MD5 digest of key,
383 # plain (undigested) key and value.
386 my ($tag, $md5, $plain_key, $value) = @_;
387 my $keys = $tag->{content};
391 # added ref() check first to avoid eval and runtime exception for every
392 # scalar value being stored. performance tweak.
393 my $is_dbm_deep = ref($value) && eval { $value->isa( 'DBM::Deep' ) };
395 my $internal_ref = $is_dbm_deep && ($value->root eq $self->root);
400 # Iterate through buckets, seeing if this is a new entry or a replace.
402 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
403 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
404 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
407 # Found empty bucket (end of list). Populate and exit loop.
411 $location = $internal_ref
412 ? $value->base_offset
413 : $self->root->{end};
415 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
416 print($fh $md5 . pack($LONG_PACK, $location) );
419 elsif ($md5 eq $key) {
421 # Found existing bucket with same key. Replace with new value.
426 $location = $value->base_offset;
427 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
428 print($fh $md5 . pack($LONG_PACK, $location) );
431 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
433 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
436 # If value is a hash, array, or raw value with equal or less size, we can
437 # reuse the same content area of the database. Otherwise, we have to create
438 # a new content area at the EOF.
441 my $r = Scalar::Util::reftype( $value ) || '';
442 if ( $r eq 'HASH' || $r eq 'ARRAY' ) {
443 $actual_length = $INDEX_SIZE;
445 # if autobless is enabled, must also take into consideration
446 # the class name, as it is stored along with key/value.
447 if ( $self->root->{autobless} ) {
448 my $value_class = Scalar::Util::blessed($value);
449 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
450 $actual_length += length($value_class);
454 else { $actual_length = length($value); }
456 if ($actual_length <= $size) {
460 $location = $self->root->{end};
461 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE, SEEK_SET);
462 print($fh pack($LONG_PACK, $location) );
470 # If this is an internal reference, return now.
471 # No need to write value or plain key
478 # If bucket didn't fit into list, split into a new index level
481 seek($fh, $tag->{ref_loc}, SEEK_SET);
482 print($fh pack($LONG_PACK, $self->root->{end}) );
484 my $index_tag = $self->_create_tag($self->root->{end}, SIG_INDEX, chr(0) x $INDEX_SIZE);
487 $keys .= $md5 . pack($LONG_PACK, 0);
489 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
490 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
492 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
493 my $num = ord(substr($key, $tag->{ch} + 1, 1));
495 if ($offsets[$num]) {
496 my $offset = $offsets[$num] + SIG_SIZE + $DATA_LENGTH_SIZE;
497 seek($fh, $offset, SEEK_SET);
499 read( $fh, $subkeys, $BUCKET_LIST_SIZE);
501 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
502 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
504 seek($fh, $offset + ($k * $BUCKET_SIZE), SEEK_SET);
505 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
511 $offsets[$num] = $self->root->{end};
512 seek($fh, $index_tag->{offset} + ($num * $LONG_SIZE), SEEK_SET);
513 print($fh pack($LONG_PACK, $self->root->{end}) );
515 my $blist_tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
517 seek($fh, $blist_tag->{offset}, SEEK_SET);
518 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
523 $location ||= $self->root->{end};
524 } # re-index bucket list
527 # Seek to content area and store signature, value and plaintext key
531 seek($fh, $location, SEEK_SET);
534 # Write signature based on content type, set content length and write actual value.
536 my $r = Scalar::Util::reftype($value) || '';
538 print($fh TYPE_HASH );
539 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
540 $content_length = $INDEX_SIZE;
542 elsif ($r eq 'ARRAY') {
543 print($fh TYPE_ARRAY );
544 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
545 $content_length = $INDEX_SIZE;
547 elsif (!defined($value)) {
548 print($fh SIG_NULL );
549 print($fh pack($DATA_LENGTH_PACK, 0) );
553 print($fh SIG_DATA );
554 print($fh pack($DATA_LENGTH_PACK, length($value)) . $value );
555 $content_length = length($value);
559 # Plain key is stored AFTER value, as keys are typically fetched less often.
561 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
564 # If value is blessed, preserve class name
566 if ( $self->root->{autobless} ) {
567 my $value_class = Scalar::Util::blessed($value);
568 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
570 # Blessed ref -- will restore later
573 print($fh pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
574 $content_length += 1;
575 $content_length += $DATA_LENGTH_SIZE + length($value_class);
579 $content_length += 1;
584 # If this is a new content area, advance EOF counter
586 if ($location == $self->root->{end}) {
587 $self->root->{end} += SIG_SIZE;
588 $self->root->{end} += $DATA_LENGTH_SIZE + $content_length;
589 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
593 # If content is a hash or array, create new child DeepDB object and
594 # pass each key or element to it.
597 my $branch = DBM::Deep->new(
599 base_offset => $location,
602 foreach my $key (keys %{$value}) {
603 #$branch->{$key} = $value->{$key};
604 $branch->STORE( $key, $value->{$key} );
607 elsif ($r eq 'ARRAY') {
608 my $branch = DBM::Deep->new(
610 base_offset => $location,
614 foreach my $element (@{$value}) {
615 #$branch->[$index] = $element;
616 $branch->STORE( $index, $element );
624 return $self->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
627 sub _get_bucket_value {
629 # Fetch single value given tag and MD5 digested key.
632 my ($tag, $md5) = @_;
633 my $keys = $tag->{content};
638 # Iterate through buckets, looking for a key match
641 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
642 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
643 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
647 # Hit end of list, no match
652 if ( $md5 ne $key ) {
657 # Found match -- seek to offset and read signature
660 seek($fh, $subloc, SEEK_SET);
661 read( $fh, $signature, SIG_SIZE);
664 # If value is a hash or array, return new DeepDB object with correct offset
666 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
667 my $obj = DBM::Deep->new(
669 base_offset => $subloc,
673 if ($self->root->{autobless}) {
675 # Skip over value and plain key to see if object needs
678 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
681 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
682 if ($size) { seek($fh, $size, SEEK_CUR); }
685 read( $fh, $bless_bit, 1);
686 if (ord($bless_bit)) {
688 # Yes, object needs to be re-blessed
691 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
692 if ($size) { read( $fh, $class_name, $size); }
693 if ($class_name) { $obj = bless( $obj, $class_name ); }
701 # Otherwise return actual value
703 elsif ($signature eq SIG_DATA) {
706 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
707 if ($size) { read( $fh, $value, $size); }
712 # Key exists, but content is null
722 # Delete single key/value pair given tag and MD5 digested key.
725 my ($tag, $md5) = @_;
726 my $keys = $tag->{content};
731 # Iterate through buckets, looking for a key match
734 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
735 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
736 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
740 # Hit end of list, no match
745 if ( $md5 ne $key ) {
750 # Matched key -- delete bucket and return
752 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
753 print($fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
754 print($fh chr(0) x $BUCKET_SIZE );
764 # Check existence of single key given tag and MD5 digested key.
767 my ($tag, $md5) = @_;
768 my $keys = $tag->{content};
771 # Iterate through buckets, looking for a key match
774 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
775 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
776 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
780 # Hit end of list, no match
785 if ( $md5 ne $key ) {
790 # Matched key -- return true
798 sub _find_bucket_list {
800 # Locate offset for bucket list, given digested key
806 # Locate offset for bucket list using digest index system
809 my $tag = $self->_load_tag($self->base_offset);
810 if (!$tag) { return; }
812 while ($tag->{signature} ne SIG_BLIST) {
813 $tag = $self->_index_lookup($tag, ord(substr($md5, $ch, 1)));
814 if (!$tag) { return; }
821 sub _traverse_index {
823 # Scan index and recursively step into deeper levels, looking for next key.
825 my ($self, $offset, $ch, $force_return_next) = @_;
826 $force_return_next = undef unless $force_return_next;
828 my $tag = $self->_load_tag( $offset );
832 if ($tag->{signature} ne SIG_BLIST) {
833 my $content = $tag->{content};
835 if ($self->{return_next}) { $start = 0; }
836 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
838 for (my $index = $start; $index < 256; $index++) {
839 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
841 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
842 if (defined($result)) { return $result; }
846 $self->{return_next} = 1;
849 elsif ($tag->{signature} eq SIG_BLIST) {
850 my $keys = $tag->{content};
851 if ($force_return_next) { $self->{return_next} = 1; }
854 # Iterate through buckets, looking for a key match
856 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
857 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
858 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
862 # End of bucket list -- return to outer loop
864 $self->{return_next} = 1;
867 elsif ($key eq $self->{prev_md5}) {
869 # Located previous key -- return next one found
871 $self->{return_next} = 1;
874 elsif ($self->{return_next}) {
876 # Seek to bucket location and skip over signature
878 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
881 # Skip over value to get to plain key
884 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
885 if ($size) { seek($fh, $size, SEEK_CUR); }
888 # Read in plain key and return as scalar
891 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
892 if ($size) { read( $fh, $plain_key, $size); }
898 $self->{return_next} = 1;
899 } # tag is a bucket list
906 # Locate next key, given digested previous one
908 my $self = $_[0]->_get_self;
910 $self->{prev_md5} = $_[1] ? $_[1] : undef;
911 $self->{return_next} = 0;
914 # If the previous key was not specifed, start at the top and
915 # return the first one found.
917 if (!$self->{prev_md5}) {
918 $self->{prev_md5} = chr(0) x $HASH_SIZE;
919 $self->{return_next} = 1;
922 return $self->_traverse_index( $self->base_offset, 0 );
927 # If db locking is set, flock() the db file. If called multiple
928 # times before unlock(), then the same number of unlocks() must
929 # be called before the lock is released.
931 my $self = $_[0]->_get_self;
933 $type = LOCK_EX unless defined $type;
935 if (!defined($self->fh)) { return; }
937 if ($self->root->{locking}) {
938 if (!$self->root->{locked}) {
939 flock($self->fh, $type);
941 # double-check file inode, in case another process
942 # has optimize()d our file while we were waiting.
943 if ((stat($self->root->{file}))[1] != $self->root->{inode}) {
944 $self->_open(); # re-open
945 flock($self->fh, $type); # re-lock
948 $self->root->{locked}++;
958 # If db locking is set, unlock the db file. See note in lock()
959 # regarding calling lock() multiple times.
961 my $self = $_[0]->_get_self;
963 if (!defined($self->fh)) { return; }
965 if ($self->root->{locking} && $self->root->{locked} > 0) {
966 $self->root->{locked}--;
967 if (!$self->root->{locked}) { flock($self->fh, LOCK_UN); }
975 #XXX These uses of ref() need verified
978 # Copy single level of keys or elements to new DB handle.
979 # Recurse for nested structures
981 my $self = $_[0]->_get_self;
984 if ($self->type eq TYPE_HASH) {
985 my $key = $self->first_key();
987 my $value = $self->get($key);
988 #XXX This doesn't work with autobless
989 if (!ref($value)) { $db_temp->{$key} = $value; }
991 my $type = $value->type;
992 if ($type eq TYPE_HASH) { $db_temp->{$key} = {}; }
993 else { $db_temp->{$key} = []; }
994 $value->_copy_node( $db_temp->{$key} );
996 $key = $self->next_key($key);
1000 my $length = $self->length();
1001 for (my $index = 0; $index < $length; $index++) {
1002 my $value = $self->get($index);
1003 if (!ref($value)) { $db_temp->[$index] = $value; }
1004 #XXX NO tests for this code
1006 my $type = $value->type;
1007 if ($type eq TYPE_HASH) { $db_temp->[$index] = {}; }
1008 else { $db_temp->[$index] = []; }
1009 $value->_copy_node( $db_temp->[$index] );
1017 # Recursively export into standard Perl hashes and arrays.
1019 my $self = $_[0]->_get_self;
1022 if ($self->type eq TYPE_HASH) { $temp = {}; }
1023 elsif ($self->type eq TYPE_ARRAY) { $temp = []; }
1026 $self->_copy_node( $temp );
1034 # Recursively import Perl hash/array structure
1036 #XXX This use of ref() seems to be ok
1037 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
1039 my $self = $_[0]->_get_self;
1042 #XXX This use of ref() seems to be ok
1043 if (!ref($struct)) {
1045 # struct is not a reference, so just import based on our type
1049 if ($self->type eq TYPE_HASH) { $struct = {@_}; }
1050 elsif ($self->type eq TYPE_ARRAY) { $struct = [@_]; }
1053 my $r = Scalar::Util::reftype($struct) || '';
1054 if ($r eq "HASH" && $self->type eq TYPE_HASH) {
1055 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
1057 elsif ($r eq "ARRAY" && $self->type eq TYPE_ARRAY) {
1058 $self->push( @$struct );
1061 return $self->_throw_error("Cannot import: type mismatch");
1069 # Rebuild entire database into new file, then move
1070 # it back on top of original.
1072 my $self = $_[0]->_get_self;
1074 #XXX Need to create a new test for this
1075 # if ($self->root->{links} > 1) {
1076 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
1079 my $db_temp = DBM::Deep->new(
1080 file => $self->root->{file} . '.tmp',
1084 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
1088 $self->_copy_node( $db_temp );
1092 # Attempt to copy user, group and permissions over to new file
1094 my @stats = stat($self->fh);
1095 my $perms = $stats[2] & 07777;
1096 my $uid = $stats[4];
1097 my $gid = $stats[5];
1098 chown( $uid, $gid, $self->root->{file} . '.tmp' );
1099 chmod( $perms, $self->root->{file} . '.tmp' );
1101 # q.v. perlport for more information on this variable
1102 if ( $^O eq 'MSWin32' ) {
1104 # Potential race condition when optmizing on Win32 with locking.
1105 # The Windows filesystem requires that the filehandle be closed
1106 # before it is overwritten with rename(). This could be redone
1113 if (!rename $self->root->{file} . '.tmp', $self->root->{file}) {
1114 unlink $self->root->{file} . '.tmp';
1116 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
1128 # Make copy of object and return
1130 my $self = $_[0]->_get_self;
1132 return DBM::Deep->new(
1133 type => $self->type,
1134 base_offset => $self->base_offset,
1140 my %is_legal_filter = map {
1143 store_key store_value
1144 fetch_key fetch_value
1149 # Setup filter function for storing or fetching the key or value
1151 my $self = $_[0]->_get_self;
1152 my $type = lc $_[1];
1153 my $func = $_[2] ? $_[2] : undef;
1155 if ( $is_legal_filter{$type} ) {
1156 $self->root->{"filter_$type"} = $func;
1170 # Get access to the root structure
1172 my $self = $_[0]->_get_self;
1173 return $self->{root};
1178 # Get access to the raw FileHandle
1180 #XXX It will be useful, though, when we split out HASH and ARRAY
1181 my $self = $_[0]->_get_self;
1182 return $self->root->{fh};
1187 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
1189 my $self = $_[0]->_get_self;
1190 return $self->{type};
1195 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
1197 my $self = $_[0]->_get_self;
1198 return $self->{base_offset};
1203 # Get last error string, or undef if no error
1206 #? ( _get_self($_[0])->{root}->{error} or undef )
1207 ? ( $_[0]->_get_self->{root}->{error} or undef )
1217 # Store error string in self
1219 my $self = $_[0]->_get_self;
1220 my $error_text = $_[1];
1222 if ( Scalar::Util::blessed $self ) {
1223 $self->root->{error} = $error_text;
1225 unless ($self->root->{debug}) {
1226 die "DBM::Deep: $error_text\n";
1229 warn "DBM::Deep: $error_text\n";
1233 die "DBM::Deep: $error_text\n";
1241 my $self = $_[0]->_get_self;
1243 undef $self->root->{error};
1248 # Precalculate index, bucket and bucket list sizes
1251 #XXX I don't like this ...
1252 set_pack() unless defined $LONG_SIZE;
1254 $INDEX_SIZE = 256 * $LONG_SIZE;
1255 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
1256 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
1261 # Set pack/unpack modes (see file header for more)
1263 my ($long_s, $long_p, $data_s, $data_p) = @_;
1265 $LONG_SIZE = $long_s ? $long_s : 4;
1266 $LONG_PACK = $long_p ? $long_p : 'N';
1268 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
1269 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
1276 # Set key digest function (default is MD5)
1278 my ($digest_func, $hash_size) = @_;
1280 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
1281 $HASH_SIZE = $hash_size ? $hash_size : 16;
1287 # tie() methods (hashes and arrays)
1292 # Store single hash key/value or array element in database.
1294 my $self = $_[0]->_get_self;
1297 # User may be storing a hash, in which case we do not want it run
1298 # through the filtering system
1299 my $value = ($self->root->{filter_store_value} && !ref($_[2]))
1300 ? $self->root->{filter_store_value}->($_[2])
1303 my $md5 = $DIGEST_FUNC->($key);
1306 # Make sure file is open
1308 if (!defined($self->fh) && !$self->_open()) {
1314 # Request exclusive lock for writing
1316 $self->lock( LOCK_EX );
1321 # If locking is enabled, set 'end' parameter again, in case another
1322 # DB instance appended to our file while we were unlocked.
1324 if ($self->root->{locking} || $self->root->{volatile}) {
1325 $self->root->{end} = (stat($fh))[7];
1329 # Locate offset for bucket list using digest index system
1331 my $tag = $self->_load_tag($self->base_offset);
1333 $tag = $self->_create_tag($self->base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1337 while ($tag->{signature} ne SIG_BLIST) {
1338 my $num = ord(substr($md5, $ch, 1));
1339 my $new_tag = $self->_index_lookup($tag, $num);
1341 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1342 seek($fh, $ref_loc, SEEK_SET);
1343 print($fh pack($LONG_PACK, $self->root->{end}) );
1345 $tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1346 $tag->{ref_loc} = $ref_loc;
1351 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1353 $tag->{ref_loc} = $ref_loc;
1360 # Add key/value to bucket list
1362 my $result = $self->_add_bucket( $tag, $md5, $key, $value );
1371 # Fetch single value or element given plain key or array index
1373 my $self = shift->_get_self;
1377 # Make sure file is open
1379 if (!defined($self->fh)) { $self->_open(); }
1381 my $md5 = $DIGEST_FUNC->($key);
1384 # Request shared lock for reading
1386 $self->lock( LOCK_SH );
1388 my $tag = $self->_find_bucket_list( $md5 );
1395 # Get value from bucket list
1397 my $result = $self->_get_bucket_value( $tag, $md5 );
1401 #XXX What is ref() checking here?
1402 #YYY Filters only apply on scalar values, so the ref check is making
1403 #YYY sure the fetched bucket is a scalar, not a child hash or array.
1404 return ($result && !ref($result) && $self->root->{filter_fetch_value})
1405 ? $self->root->{filter_fetch_value}->($result)
1411 # Delete single key/value pair or element given plain key or array index
1413 my $self = $_[0]->_get_self;
1416 my $md5 = $DIGEST_FUNC->($key);
1419 # Make sure file is open
1421 if (!defined($self->fh)) { $self->_open(); }
1424 # Request exclusive lock for writing
1426 $self->lock( LOCK_EX );
1428 my $tag = $self->_find_bucket_list( $md5 );
1437 my $value = $self->_get_bucket_value( $tag, $md5 );
1438 if ($value && !ref($value) && $self->root->{filter_fetch_value}) {
1439 $value = $self->root->{filter_fetch_value}->($value);
1442 my $result = $self->_delete_bucket( $tag, $md5 );
1445 # If this object is an array and the key deleted was on the end of the stack,
1446 # decrement the length variable.
1456 # Check if a single key or element exists given plain key or array index
1458 my $self = $_[0]->_get_self;
1461 my $md5 = $DIGEST_FUNC->($key);
1464 # Make sure file is open
1466 if (!defined($self->fh)) { $self->_open(); }
1469 # Request shared lock for reading
1471 $self->lock( LOCK_SH );
1473 my $tag = $self->_find_bucket_list( $md5 );
1476 # For some reason, the built-in exists() function returns '' for false
1484 # Check if bucket exists and return 1 or ''
1486 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1495 # Clear all keys from hash, or all elements from array.
1497 my $self = $_[0]->_get_self;
1500 # Make sure file is open
1502 if (!defined($self->fh)) { $self->_open(); }
1505 # Request exclusive lock for writing
1507 $self->lock( LOCK_EX );
1511 seek($fh, $self->base_offset, SEEK_SET);
1517 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
1525 # Public method aliases
1527 sub put { (shift)->STORE( @_ ) }
1528 sub store { (shift)->STORE( @_ ) }
1529 sub get { (shift)->FETCH( @_ ) }
1530 sub fetch { (shift)->FETCH( @_ ) }
1531 sub delete { (shift)->DELETE( @_ ) }
1532 sub exists { (shift)->EXISTS( @_ ) }
1533 sub clear { (shift)->CLEAR( @_ ) }
1535 package DBM::Deep::_::Root;
1550 filter_store_key => undef,
1551 filter_store_value => undef,
1552 filter_fetch_key => undef,
1553 filter_fetch_value => undef,
1564 return unless $self;
1566 close $self->{fh} if $self->{fh};
1577 DBM::Deep - A pure perl multi-level hash/array DBM
1582 my $db = DBM::Deep->new( "foo.db" );
1584 $db->{key} = 'value'; # tie() style
1587 $db->put('key', 'value'); # OO style
1588 print $db->get('key');
1590 # true multi-level support
1591 $db->{my_complex} = [
1592 'hello', { perl => 'rules' },
1597 A unique flat-file database module, written in pure perl. True
1598 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1599 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1600 handle millions of keys and unlimited hash levels without significant
1601 slow-down. Written from the ground-up in pure perl -- this is NOT a
1602 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1603 Mac OS X and Windows.
1607 Hopefully you are using CPAN's excellent Perl module, which will download
1608 and install the module for you. If not, get the tarball, and run these
1620 Construction can be done OO-style (which is the recommended way), or using
1621 Perl's tie() function. Both are examined here.
1623 =head2 OO CONSTRUCTION
1625 The recommended way to construct a DBM::Deep object is to use the new()
1626 method, which gets you a blessed, tied hash or array reference.
1628 my $db = DBM::Deep->new( "foo.db" );
1630 This opens a new database handle, mapped to the file "foo.db". If this
1631 file does not exist, it will automatically be created. DB files are
1632 opened in "r+" (read/write) mode, and the type of object returned is a
1633 hash, unless otherwise specified (see L<OPTIONS> below).
1637 You can pass a number of options to the constructor to specify things like
1638 locking, autoflush, etc. This is done by passing an inline hash:
1640 my $db = DBM::Deep->new(
1646 Notice that the filename is now specified I<inside> the hash with
1647 the "file" parameter, as opposed to being the sole argument to the
1648 constructor. This is required if any options are specified.
1649 See L<OPTIONS> below for the complete list.
1653 You can also start with an array instead of a hash. For this, you must
1654 specify the C<type> parameter:
1656 my $db = DBM::Deep->new(
1658 type => DBM::Deep->TYPE_ARRAY
1661 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1662 a new DB file. If you create a DBM::Deep object with an existing file, the
1663 C<type> will be loaded from the file header, and ignored if it is passed
1666 =head2 TIE CONSTRUCTION
1668 Alternatively, you can create a DBM::Deep handle by using Perl's built-in
1669 tie() function. This is not ideal, because you get only a basic, tied hash
1670 (or array) which is not blessed, so you can't call any functions on it.
1673 tie %hash, "DBM::Deep", "foo.db";
1676 tie @array, "DBM::Deep", "bar.db";
1678 As with the OO constructor, you can replace the DB filename parameter with
1679 a hash containing one or more options (see L<OPTIONS> just below for the
1682 tie %hash, "DBM::Deep", {
1690 There are a number of options that can be passed in when constructing your
1691 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1697 Filename of the DB file to link the handle to. You can pass a full absolute
1698 filesystem path, partial path, or a plain filename if the file is in the
1699 current working directory. This is a required parameter.
1703 File open mode (read-only, read-write, etc.) string passed to Perl's FileHandle
1704 module. This is an optional parameter, and defaults to "r+" (read/write).
1705 B<Note:> If the default (r+) mode is selected, the file will also be auto-
1706 created if it doesn't exist.
1710 This parameter specifies what type of object to create, a hash or array. Use
1711 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1712 This only takes effect when beginning a new file. This is an optional
1713 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1717 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1718 function to lock the database in exclusive mode for writes, and shared mode for
1719 reads. Pass any true value to enable. This affects the base DB handle I<and
1720 any child hashes or arrays> that use the same DB file. This is an optional
1721 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1725 Specifies whether autoflush is to be enabled on the underlying FileHandle.
1726 This obviously slows down write operations, but is required if you may have
1727 multiple processes accessing the same DB file (also consider enable I<locking>
1728 or at least I<volatile>). Pass any true value to enable. This is an optional
1729 parameter, and defaults to 0 (disabled).
1733 If I<volatile> mode is enabled, DBM::Deep will stat() the DB file before each
1734 STORE() operation. This is required if an outside force may change the size of
1735 the file between transactions. Locking also implicitly enables volatile. This
1736 is useful if you want to use a different locking system or write your own. Pass
1737 any true value to enable. This is an optional parameter, and defaults to 0
1742 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1743 restore them when fetched. This is an B<experimental> feature, and does have
1744 side-effects. Basically, when hashes are re-blessed into their original
1745 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1746 able to call any DBM::Deep methods on them. You have been warned.
1747 This is an optional parameter, and defaults to 0 (disabled).
1751 See L<FILTERS> below.
1755 Setting I<debug> mode will make all errors non-fatal, dump them out to
1756 STDERR, and continue on. This is for debugging purposes only, and probably
1757 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1761 Instead of passing a file path, you can instead pass a handle to an pre-opened
1762 filehandle. Note: Beware of using the magick *DATA handle, as this actually
1763 contains your entire Perl script, as well as the data following the __DATA__
1764 marker. This will not work, because DBM::Deep uses absolute seek()s into the
1765 file. Instead, consider reading *DATA into an IO::Scalar handle, then passing
1766 in that. Also please note optimize() will NOT work when passing in only a
1767 handle. Pass in a real filename in order to use optimize().
1771 =head1 TIE INTERFACE
1773 With DBM::Deep you can access your databases using Perl's standard hash/array
1774 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can treat
1775 them as such. DBM::Deep will intercept all reads/writes and direct them to the right
1776 place -- the DB file. This has nothing to do with the L<TIE CONSTRUCTION>
1777 section above. This simply tells you how to use DBM::Deep using regular hashes
1778 and arrays, rather than calling functions like C<get()> and C<put()> (although those
1779 work too). It is entirely up to you how to want to access your databases.
1783 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1784 or even nested hashes (or arrays) using standard Perl syntax:
1786 my $db = DBM::Deep->new( "foo.db" );
1788 $db->{mykey} = "myvalue";
1790 $db->{myhash}->{subkey} = "subvalue";
1792 print $db->{myhash}->{subkey} . "\n";
1794 You can even step through hash keys using the normal Perl C<keys()> function:
1796 foreach my $key (keys %$db) {
1797 print "$key: " . $db->{$key} . "\n";
1800 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1801 pushes them onto an array, all before the loop even begins. If you have an
1802 extra large hash, this may exhaust Perl's memory. Instead, consider using
1803 Perl's C<each()> function, which pulls keys/values one at a time, using very
1806 while (my ($key, $value) = each %$db) {
1807 print "$key: $value\n";
1810 Please note that when using C<each()>, you should always pass a direct
1811 hash reference, not a lookup. Meaning, you should B<never> do this:
1814 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1816 This causes an infinite loop, because for each iteration, Perl is calling
1817 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1818 it effectively keeps returning the first key over and over again. Instead,
1819 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1823 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1824 reference. This includes inserting, removing and manipulating elements,
1825 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1826 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1827 or simply be a nested array reference inside a hash. Example:
1829 my $db = DBM::Deep->new(
1830 file => "foo-array.db",
1831 type => DBM::Deep->TYPE_ARRAY
1835 push @$db, "bar", "baz";
1836 unshift @$db, "bah";
1838 my $last_elem = pop @$db; # baz
1839 my $first_elem = shift @$db; # bah
1840 my $second_elem = $db->[1]; # bar
1842 my $num_elements = scalar @$db;
1846 In addition to the I<tie()> interface, you can also use a standard OO interface
1847 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1848 array) has its own methods, but both types share the following common methods:
1849 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1855 Stores a new hash key/value pair, or sets an array element value. Takes two
1856 arguments, the hash key or array index, and the new value. The value can be
1857 a scalar, hash ref or array ref. Returns true on success, false on failure.
1859 $db->put("foo", "bar"); # for hashes
1860 $db->put(1, "bar"); # for arrays
1864 Fetches the value of a hash key or array element. Takes one argument: the hash
1865 key or array index. Returns a scalar, hash ref or array ref, depending on the
1868 my $value = $db->get("foo"); # for hashes
1869 my $value = $db->get(1); # for arrays
1873 Checks if a hash key or array index exists. Takes one argument: the hash key
1874 or array index. Returns true if it exists, false if not.
1876 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1877 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1881 Deletes one hash key/value pair or array element. Takes one argument: the hash
1882 key or array index. Returns true on success, false if not found. For arrays,
1883 the remaining elements located after the deleted element are NOT moved over.
1884 The deleted element is essentially just undefined, which is exactly how Perl's
1885 internal arrays work. Please note that the space occupied by the deleted
1886 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1887 below for details and workarounds.
1889 $db->delete("foo"); # for hashes
1890 $db->delete(1); # for arrays
1894 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1895 value. Please note that the space occupied by the deleted keys/values or
1896 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1897 details and workarounds.
1899 $db->clear(); # hashes or arrays
1905 For hashes, DBM::Deep supports all the common methods described above, and the
1906 following additional methods: C<first_key()> and C<next_key()>.
1912 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1913 fetched in an undefined order (which appears random). Takes no arguments,
1914 returns the key as a scalar value.
1916 my $key = $db->first_key();
1920 Returns the "next" key in the hash, given the previous one as the sole argument.
1921 Returns undef if there are no more keys to be fetched.
1923 $key = $db->next_key($key);
1927 Here are some examples of using hashes:
1929 my $db = DBM::Deep->new( "foo.db" );
1931 $db->put("foo", "bar");
1932 print "foo: " . $db->get("foo") . "\n";
1934 $db->put("baz", {}); # new child hash ref
1935 $db->get("baz")->put("buz", "biz");
1936 print "buz: " . $db->get("baz")->get("buz") . "\n";
1938 my $key = $db->first_key();
1940 print "$key: " . $db->get($key) . "\n";
1941 $key = $db->next_key($key);
1944 if ($db->exists("foo")) { $db->delete("foo"); }
1948 For arrays, DBM::Deep supports all the common methods described above, and the
1949 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1950 C<unshift()> and C<splice()>.
1956 Returns the number of elements in the array. Takes no arguments.
1958 my $len = $db->length();
1962 Adds one or more elements onto the end of the array. Accepts scalars, hash
1963 refs or array refs. No return value.
1965 $db->push("foo", "bar", {});
1969 Fetches the last element in the array, and deletes it. Takes no arguments.
1970 Returns undef if array is empty. Returns the element value.
1972 my $elem = $db->pop();
1976 Fetches the first element in the array, deletes it, then shifts all the
1977 remaining elements over to take up the space. Returns the element value. This
1978 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1981 my $elem = $db->shift();
1985 Inserts one or more elements onto the beginning of the array, shifting all
1986 existing elements over to make room. Accepts scalars, hash refs or array refs.
1987 No return value. This method is not recommended with large arrays -- see
1988 <LARGE ARRAYS> below for details.
1990 $db->unshift("foo", "bar", {});
1994 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1995 -f splice> for usage -- it is too complicated to document here. This method is
1996 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
2000 Here are some examples of using arrays:
2002 my $db = DBM::Deep->new(
2004 type => DBM::Deep->TYPE_ARRAY
2007 $db->push("bar", "baz");
2008 $db->unshift("foo");
2011 my $len = $db->length();
2012 print "length: $len\n"; # 4
2014 for (my $k=0; $k<$len; $k++) {
2015 print "$k: " . $db->get($k) . "\n";
2018 $db->splice(1, 2, "biz", "baf");
2020 while (my $elem = shift @$db) {
2021 print "shifted: $elem\n";
2026 Enable automatic file locking by passing a true value to the C<locking>
2027 parameter when constructing your DBM::Deep object (see L<SETUP> above).
2029 my $db = DBM::Deep->new(
2034 This causes DBM::Deep to C<flock()> the underlying FileHandle object with exclusive
2035 mode for writes, and shared mode for reads. This is required if you have
2036 multiple processes accessing the same database file, to avoid file corruption.
2037 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
2038 NFS> below for more.
2040 =head2 EXPLICIT LOCKING
2042 You can explicitly lock a database, so it remains locked for multiple
2043 transactions. This is done by calling the C<lock()> method, and passing an
2044 optional lock mode argument (defaults to exclusive mode). This is particularly
2045 useful for things like counters, where the current value needs to be fetched,
2046 then incremented, then stored again.
2049 my $counter = $db->get("counter");
2051 $db->put("counter", $counter);
2060 You can pass C<lock()> an optional argument, which specifies which mode to use
2061 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2062 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2063 same as the constants defined in Perl's C<Fcntl> module.
2065 $db->lock( DBM::Deep->LOCK_SH );
2069 If you want to implement your own file locking scheme, be sure to create your
2070 DBM::Deep objects setting the C<volatile> option to true. This hints to DBM::Deep
2071 that the DB file may change between transactions. See L<LOW-LEVEL ACCESS>
2074 =head1 IMPORTING/EXPORTING
2076 You can import existing complex structures by calling the C<import()> method,
2077 and export an entire database into an in-memory structure using the C<export()>
2078 method. Both are examined here.
2082 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2083 walking the structure and adding keys/elements to the database as you go,
2084 simply pass a reference to the C<import()> method. This recursively adds
2085 everything to an existing DBM::Deep object for you. Here is an example:
2090 array1 => [ "elem0", "elem1", "elem2" ],
2092 subkey1 => "subvalue1",
2093 subkey2 => "subvalue2"
2097 my $db = DBM::Deep->new( "foo.db" );
2098 $db->import( $struct );
2100 print $db->{key1} . "\n"; # prints "value1"
2102 This recursively imports the entire C<$struct> object into C<$db>, including
2103 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2104 keys are merged with the existing ones, replacing if they already exist.
2105 The C<import()> method can be called on any database level (not just the base
2106 level), and works with both hash and array DB types.
2110 B<Note:> Make sure your existing structure has no circular references in it.
2111 These will cause an infinite loop when importing.
2115 Calling the C<export()> method on an existing DBM::Deep object will return
2116 a reference to a new in-memory copy of the database. The export is done
2117 recursively, so all nested hashes/arrays are all exported to standard Perl
2118 objects. Here is an example:
2120 my $db = DBM::Deep->new( "foo.db" );
2122 $db->{key1} = "value1";
2123 $db->{key2} = "value2";
2125 $db->{hash1}->{subkey1} = "subvalue1";
2126 $db->{hash1}->{subkey2} = "subvalue2";
2128 my $struct = $db->export();
2130 print $struct->{key1} . "\n"; # prints "value1"
2132 This makes a complete copy of the database in memory, and returns a reference
2133 to it. The C<export()> method can be called on any database level (not just
2134 the base level), and works with both hash and array DB types. Be careful of
2135 large databases -- you can store a lot more data in a DBM::Deep object than an
2136 in-memory Perl structure.
2140 B<Note:> Make sure your database has no circular references in it.
2141 These will cause an infinite loop when exporting.
2145 DBM::Deep has a number of hooks where you can specify your own Perl function
2146 to perform filtering on incoming or outgoing data. This is a perfect
2147 way to extend the engine, and implement things like real-time compression or
2148 encryption. Filtering applies to the base DB level, and all child hashes /
2149 arrays. Filter hooks can be specified when your DBM::Deep object is first
2150 constructed, or by calling the C<set_filter()> method at any time. There are
2151 four available filter hooks, described below:
2155 =item * filter_store_key
2157 This filter is called whenever a hash key is stored. It
2158 is passed the incoming key, and expected to return a transformed key.
2160 =item * filter_store_value
2162 This filter is called whenever a hash key or array element is stored. It
2163 is passed the incoming value, and expected to return a transformed value.
2165 =item * filter_fetch_key
2167 This filter is called whenever a hash key is fetched (i.e. via
2168 C<first_key()> or C<next_key()>). It is passed the transformed key,
2169 and expected to return the plain key.
2171 =item * filter_fetch_value
2173 This filter is called whenever a hash key or array element is fetched.
2174 It is passed the transformed value, and expected to return the plain value.
2178 Here are the two ways to setup a filter hook:
2180 my $db = DBM::Deep->new(
2182 filter_store_value => \&my_filter_store,
2183 filter_fetch_value => \&my_filter_fetch
2188 $db->set_filter( "filter_store_value", \&my_filter_store );
2189 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2191 Your filter function will be called only when dealing with SCALAR keys or
2192 values. When nested hashes and arrays are being stored/fetched, filtering
2193 is bypassed. Filters are called as static functions, passed a single SCALAR
2194 argument, and expected to return a single SCALAR value. If you want to
2195 remove a filter, set the function reference to C<undef>:
2197 $db->set_filter( "filter_store_value", undef );
2199 =head2 REAL-TIME ENCRYPTION EXAMPLE
2201 Here is a working example that uses the I<Crypt::Blowfish> module to
2202 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2203 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2204 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2207 use Crypt::Blowfish;
2210 my $cipher = Crypt::CBC->new({
2211 'key' => 'my secret key',
2212 'cipher' => 'Blowfish',
2214 'regenerate_key' => 0,
2215 'padding' => 'space',
2219 my $db = DBM::Deep->new(
2220 file => "foo-encrypt.db",
2221 filter_store_key => \&my_encrypt,
2222 filter_store_value => \&my_encrypt,
2223 filter_fetch_key => \&my_decrypt,
2224 filter_fetch_value => \&my_decrypt,
2227 $db->{key1} = "value1";
2228 $db->{key2} = "value2";
2229 print "key1: " . $db->{key1} . "\n";
2230 print "key2: " . $db->{key2} . "\n";
2236 return $cipher->encrypt( $_[0] );
2239 return $cipher->decrypt( $_[0] );
2242 =head2 REAL-TIME COMPRESSION EXAMPLE
2244 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2245 compression / decompression of keys & values with DBM::Deep Filters.
2246 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2247 more on I<Compress::Zlib>.
2252 my $db = DBM::Deep->new(
2253 file => "foo-compress.db",
2254 filter_store_key => \&my_compress,
2255 filter_store_value => \&my_compress,
2256 filter_fetch_key => \&my_decompress,
2257 filter_fetch_value => \&my_decompress,
2260 $db->{key1} = "value1";
2261 $db->{key2} = "value2";
2262 print "key1: " . $db->{key1} . "\n";
2263 print "key2: " . $db->{key2} . "\n";
2269 return Compress::Zlib::memGzip( $_[0] ) ;
2272 return Compress::Zlib::memGunzip( $_[0] ) ;
2275 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2276 actually numerical index numbers, and are not filtered.
2278 =head1 ERROR HANDLING
2280 Most DBM::Deep methods return a true value for success, and call die() on
2281 failure. You can wrap calls in an eval block to catch the die. Also, the
2282 actual error message is stored in an internal scalar, which can be fetched by
2283 calling the C<error()> method.
2285 my $db = DBM::Deep->new( "foo.db" ); # create hash
2286 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2288 print $db->error(); # prints error message
2290 You can then call C<clear_error()> to clear the current error state.
2294 If you set the C<debug> option to true when creating your DBM::Deep object,
2295 all errors are considered NON-FATAL, and dumped to STDERR. This is only
2296 for debugging purposes.
2298 =head1 LARGEFILE SUPPORT
2300 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2301 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2302 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2303 by calling the static C<set_pack()> method before you do anything else.
2305 DBM::Deep::set_pack(8, 'Q');
2307 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2308 instead of 32-bit longs. After setting these values your DB files have a
2309 theoretical maximum size of 16 XB (exabytes).
2313 B<Note:> Changing these values will B<NOT> work for existing database files.
2314 Only change this for new files, and make sure it stays set consistently
2315 throughout the file's life. If you do set these values, you can no longer
2316 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2317 back to 32-bit mode.
2321 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2322 only a 32-bit Perl. However, I have received user reports that this does
2325 =head1 LOW-LEVEL ACCESS
2327 If you require low-level access to the underlying FileHandle that DBM::Deep uses,
2328 you can call the C<fh()> method, which returns the handle:
2332 This method can be called on the root level of the datbase, or any child
2333 hashes or arrays. All levels share a I<root> structure, which contains things
2334 like the FileHandle, a reference counter, and all your options you specified
2335 when you created the object. You can get access to this root structure by
2336 calling the C<root()> method.
2338 my $root = $db->root();
2340 This is useful for changing options after the object has already been created,
2341 such as enabling/disabling locking, volatile or debug modes. You can also
2342 store your own temporary user data in this structure (be wary of name
2343 collision), which is then accessible from any child hash or array.
2345 =head1 CUSTOM DIGEST ALGORITHM
2347 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2348 keys. However you can override this, and use another algorithm (such as SHA-256)
2349 or even write your own. But please note that DBM::Deep currently expects zero
2350 collisions, so your algorithm has to be I<perfect>, so to speak.
2351 Collision detection may be introduced in a later version.
2355 You can specify a custom digest algorithm by calling the static C<set_digest()>
2356 function, passing a reference to a subroutine, and the length of the algorithm's
2357 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2358 objects. Here is a working example that uses a 256-bit hash from the
2359 I<Digest::SHA256> module. Please see
2360 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2365 my $context = Digest::SHA256::new(256);
2367 DBM::Deep::set_digest( \&my_digest, 32 );
2369 my $db = DBM::Deep->new( "foo-sha.db" );
2371 $db->{key1} = "value1";
2372 $db->{key2} = "value2";
2373 print "key1: " . $db->{key1} . "\n";
2374 print "key2: " . $db->{key2} . "\n";
2380 return substr( $context->hash($_[0]), 0, 32 );
2383 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2384 of bytes you specify in the C<set_digest()> function (in this case 32).
2386 =head1 CIRCULAR REFERENCES
2388 DBM::Deep has B<experimental> support for circular references. Meaning you
2389 can have a nested hash key or array element that points to a parent object.
2390 This relationship is stored in the DB file, and is preserved between sessions.
2393 my $db = DBM::Deep->new( "foo.db" );
2396 $db->{circle} = $db; # ref to self
2398 print $db->{foo} . "\n"; # prints "foo"
2399 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2401 One catch is, passing the object to a function that recursively walks the
2402 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2403 C<export()> methods) will result in an infinite loop. The other catch is,
2404 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2405 or C<next_key()> methods), you will get the I<target object's key>, not the
2406 ref's key. This gets even more interesting with the above example, where
2407 the I<circle> key points to the base DB object, which technically doesn't
2408 have a key. So I made DBM::Deep return "[base]" as the key name in that
2411 =head1 CAVEATS / ISSUES / BUGS
2413 This section describes all the known issues with DBM::Deep. It you have found
2414 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2416 =head2 UNUSED SPACE RECOVERY
2418 One major caveat with DBM::Deep is that space occupied by existing keys and
2419 values is not recovered when they are deleted. Meaning if you keep deleting
2420 and adding new keys, your file will continuously grow. I am working on this,
2421 but in the meantime you can call the built-in C<optimize()> method from time to
2422 time (perhaps in a crontab or something) to recover all your unused space.
2424 $db->optimize(); # returns true on success
2426 This rebuilds the ENTIRE database into a new file, then moves it on top of
2427 the original. The new file will have no unused space, thus it will take up as
2428 little disk space as possible. Please note that this operation can take
2429 a long time for large files, and you need enough disk space to temporarily hold
2430 2 copies of your DB file. The temporary file is created in the same directory
2431 as the original, named with a ".tmp" extension, and is deleted when the
2432 operation completes. Oh, and if locking is enabled, the DB is automatically
2433 locked for the entire duration of the copy.
2437 B<WARNING:> Only call optimize() on the top-level node of the database, and
2438 make sure there are no child references lying around. DBM::Deep keeps a reference
2439 counter, and if it is greater than 1, optimize() will abort and return undef.
2441 =head2 AUTOVIVIFICATION
2443 Unfortunately, autovivification doesn't work with tied hashes. This appears to
2444 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
2445 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
2446 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
2447 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
2450 $db->{foo}->{bar} = "hello";
2452 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
2453 being an empty hash. Try this instead, which works fine:
2455 $db->{foo} = { bar => "hello" };
2457 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
2458 the execution path, and Perl indeed passes an empty hash to the STORE() method.
2459 Probably a bug in Perl.
2461 =head2 FILE CORRUPTION
2463 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2464 for a 32-bit signature when opened, but other corruption in files can cause
2465 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2466 stuck in an infinite loop depending on the level of corruption. File write
2467 operations are not checked for failure (for speed), so if you happen to run
2468 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2469 be addressed in a later version of DBM::Deep.
2473 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2474 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2475 about setting up your NFS server with a locking daemon, then using lockf() to
2476 lock your files, but your milage may vary there as well. From what I
2477 understand, there is no real way to do it. However, if you need access to the
2478 underlying FileHandle in DBM::Deep for using some other kind of locking scheme like
2479 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2481 =head2 COPYING OBJECTS
2483 Beware of copying tied objects in Perl. Very strange things can happen.
2484 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2485 returns a new, blessed, tied hash or array to the same level in the DB.
2487 my $copy = $db->clone();
2491 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2492 These functions cause every element in the array to move, which can be murder
2493 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2494 a different location. This may be addressed in a later version.
2498 This section discusses DBM::Deep's speed and memory usage.
2502 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2503 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2504 multi-level hash/array support, and cross-platform FTPable files. Even so,
2505 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2506 with huge databases. Here is some test data:
2508 Adding 1,000,000 keys to new DB file...
2510 At 100 keys, avg. speed is 2,703 keys/sec
2511 At 200 keys, avg. speed is 2,642 keys/sec
2512 At 300 keys, avg. speed is 2,598 keys/sec
2513 At 400 keys, avg. speed is 2,578 keys/sec
2514 At 500 keys, avg. speed is 2,722 keys/sec
2515 At 600 keys, avg. speed is 2,628 keys/sec
2516 At 700 keys, avg. speed is 2,700 keys/sec
2517 At 800 keys, avg. speed is 2,607 keys/sec
2518 At 900 keys, avg. speed is 2,190 keys/sec
2519 At 1,000 keys, avg. speed is 2,570 keys/sec
2520 At 2,000 keys, avg. speed is 2,417 keys/sec
2521 At 3,000 keys, avg. speed is 1,982 keys/sec
2522 At 4,000 keys, avg. speed is 1,568 keys/sec
2523 At 5,000 keys, avg. speed is 1,533 keys/sec
2524 At 6,000 keys, avg. speed is 1,787 keys/sec
2525 At 7,000 keys, avg. speed is 1,977 keys/sec
2526 At 8,000 keys, avg. speed is 2,028 keys/sec
2527 At 9,000 keys, avg. speed is 2,077 keys/sec
2528 At 10,000 keys, avg. speed is 2,031 keys/sec
2529 At 20,000 keys, avg. speed is 1,970 keys/sec
2530 At 30,000 keys, avg. speed is 2,050 keys/sec
2531 At 40,000 keys, avg. speed is 2,073 keys/sec
2532 At 50,000 keys, avg. speed is 1,973 keys/sec
2533 At 60,000 keys, avg. speed is 1,914 keys/sec
2534 At 70,000 keys, avg. speed is 2,091 keys/sec
2535 At 80,000 keys, avg. speed is 2,103 keys/sec
2536 At 90,000 keys, avg. speed is 1,886 keys/sec
2537 At 100,000 keys, avg. speed is 1,970 keys/sec
2538 At 200,000 keys, avg. speed is 2,053 keys/sec
2539 At 300,000 keys, avg. speed is 1,697 keys/sec
2540 At 400,000 keys, avg. speed is 1,838 keys/sec
2541 At 500,000 keys, avg. speed is 1,941 keys/sec
2542 At 600,000 keys, avg. speed is 1,930 keys/sec
2543 At 700,000 keys, avg. speed is 1,735 keys/sec
2544 At 800,000 keys, avg. speed is 1,795 keys/sec
2545 At 900,000 keys, avg. speed is 1,221 keys/sec
2546 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2548 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2549 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2550 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2551 Run time was 12 min 3 sec.
2555 One of the great things about DBM::Deep is that it uses very little memory.
2556 Even with huge databases (1,000,000+ keys) you will not see much increased
2557 memory on your process. DBM::Deep relies solely on the filesystem for storing
2558 and fetching data. Here is output from I</usr/bin/top> before even opening a
2561 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2562 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2564 Basically the process is taking 2,716K of memory. And here is the same
2565 process after storing and fetching 1,000,000 keys:
2567 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2568 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2570 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2571 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2573 =head1 DB FILE FORMAT
2575 In case you were interested in the underlying DB file format, it is documented
2576 here in this section. You don't need to know this to use the module, it's just
2577 included for reference.
2581 DBM::Deep files always start with a 32-bit signature to identify the file type.
2582 This is at offset 0. The signature is "DPDB" in network byte order. This is
2583 checked when the file is opened.
2587 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2588 has a standard header containing the type of data, the length of data, and then
2589 the data itself. The type is a single character (1 byte), the length is a
2590 32-bit unsigned long in network byte order, and the data is, well, the data.
2591 Here is how it unfolds:
2595 Immediately after the 32-bit file signature is the I<Master Index> record.
2596 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2597 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2598 depending on how the DBM::Deep object was constructed.
2602 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2603 number). The first 8-bit char of the MD5 signature is the offset into the
2604 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2605 index element is a file offset of the next tag for the key/element in question,
2606 which is usually a I<Bucket List> tag (see below).
2610 The next tag I<could> be another index, depending on how many keys/elements
2611 exist. See L<RE-INDEXING> below for details.
2615 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2616 file offsets to where the actual data is stored. It starts with a standard
2617 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2618 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2619 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2620 When the list fills up, a I<Re-Index> operation is performed (See
2621 L<RE-INDEXING> below).
2625 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2626 index/value pair (in array mode). It starts with a standard tag header with
2627 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2628 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2629 header. The size reported in the tag header is only for the value, but then,
2630 just after the value is another size (32-bit unsigned long) and then the plain
2631 key itself. Since the value is likely to be fetched more often than the plain
2632 key, I figured it would be I<slightly> faster to store the value first.
2636 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2637 record for the nested structure, where the process begins all over again.
2641 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2642 exhausted. Then, when another key/element comes in, the list is converted to a
2643 new index record. However, this index will look at the next char in the MD5
2644 hash, and arrange new Bucket List pointers accordingly. This process is called
2645 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2646 17 (16 + new one) keys/elements are removed from the old Bucket List and
2647 inserted into the new index. Several new Bucket Lists are created in the
2648 process, as a new MD5 char from the key is being examined (it is unlikely that
2649 the keys will all share the same next char of their MD5s).
2653 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2654 when the Bucket Lists will turn into indexes, but the first round tends to
2655 happen right around 4,000 keys. You will see a I<slight> decrease in
2656 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2657 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2658 right around 900,000 keys. This process can continue nearly indefinitely --
2659 right up until the point the I<MD5> signatures start colliding with each other,
2660 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2661 getting struck by lightning while you are walking to cash in your tickets.
2662 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2663 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2664 this is 340 unodecillion, but don't quote me).
2668 When a new key/element is stored, the key (or index number) is first ran through
2669 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2670 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2671 for the first char of the signature (in this case I<b>). If it does not exist,
2672 a new I<Bucket List> is created for our key (and the next 15 future keys that
2673 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2674 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2675 this point, unless we are replacing an existing I<Bucket>), where the actual
2676 data will be stored.
2680 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2681 (or index number), then walking along the indexes. If there are enough
2682 keys/elements in this DB level, there might be nested indexes, each linked to
2683 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2684 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2685 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2686 plain key are stored.
2690 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2691 methods. In this process the indexes are walked systematically, and each key
2692 fetched in increasing MD5 order (which is why it appears random). Once the
2693 I<Bucket> is found, the value is skipped the plain key returned instead.
2694 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2695 alphabetically sorted. This only happens on an index-level -- as soon as the
2696 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2697 so it's pretty much undefined how the keys will come out -- just like Perl's
2700 =head1 CODE COVERAGE
2702 I use B<Devel::Cover> to test the code coverage of my tests, below is the B<Devel::Cover>
2703 report on this module's test suite.
2705 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2706 File stmt bran cond sub pod time total
2707 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2708 blib/lib/DBM/Deep.pm 93.7 82.5 71.9 96.5 25.9 82.8 87.9
2709 blib/lib/DBM/Deep/Array.pm 98.8 88.0 90.9 100.0 n/a 12.8 96.3
2710 blib/lib/DBM/Deep/Hash.pm 95.2 80.0 100.0 100.0 n/a 4.4 92.3
2711 Total 94.8 83.2 76.5 97.6 25.9 100.0 89.7
2712 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2716 Joseph Huckaby, L<jhuckaby@cpan.org>
2717 Rob Kinyon, L<rkinyon@cpan.org>
2719 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2723 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2724 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2728 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
2729 This is free software, you may use it and distribute it under the
2730 same terms as Perl itself.