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) 2002-2006 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 );
39 $VERSION = q(0.981_02);
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 () { SIG_HASH }
104 sub TYPE_ARRAY () { SIG_ARRAY }
105 sub TYPE_SCALAR () { SIG_SCALAR }
111 if (scalar(@_) > 1) {
113 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
117 elsif ( ref $_[0] ) {
118 unless ( eval { local $SIG{'__DIE__'}; %{$_[0]} || 1 } ) {
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
169 base_offset => length(SIG_FILE),
172 foreach my $param ( keys %$self ) {
173 next unless exists $args->{$param};
174 $self->{$param} = delete $args->{$param}
177 # locking implicitly enables autoflush
178 if ($args->{locking}) { $args->{autoflush} = 1; }
180 $self->{root} = exists $args->{root}
182 : DBM::Deep::_::Root->new( $args );
184 if (!defined($self->_fh)) { $self->_open(); }
191 require DBM::Deep::Hash;
192 return DBM::Deep::Hash->TIEHASH( @_ );
197 require DBM::Deep::Array;
198 return DBM::Deep::Array->TIEARRAY( @_ );
201 #XXX Unneeded now ...
207 # Open a fh to the database, create if nonexistent.
208 # Make sure file signature matches DBM::Deep spec.
210 my $self = $_[0]->_get_self;
212 if (defined($self->_fh)) { $self->_close(); }
215 local $SIG{'__DIE__'};
216 # Theoretically, adding O_BINARY should remove the need for the binmode
217 # Of course, testing it is going to be ... interesting.
218 my $flags = O_RDWR | O_CREAT | O_BINARY;
221 sysopen( $fh, $self->_root->{file}, $flags )
223 $self->_root->{fh} = $fh;
224 }; if ($@ ) { $self->_throw_error( "Received error: $@\n" ); }
225 if (! defined($self->_fh)) {
226 return $self->_throw_error("Cannot sysopen file: " . $self->_root->{file} . ": $!");
231 #XXX Can we remove this by using the right sysopen() flags?
232 # Maybe ... q.v. above
233 binmode $fh; # for win32
235 if ($self->_root->{autoflush}) {
236 my $old = select $fh;
241 seek($fh, 0 + $self->_root->{file_offset}, SEEK_SET);
244 my $bytes_read = read( $fh, $signature, length(SIG_FILE));
247 # File is empty -- write signature and master index
250 if ( my $afh = $self->_root->{audit_fh} ) {
251 flock( $afh, LOCK_EX );
252 print( $afh "# Database created on " . localtime(time) . $/ );
253 flock( $afh, LOCK_UN );
256 seek($fh, 0 + $self->_root->{file_offset}, SEEK_SET);
257 print( $fh 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 );
263 # Flush the filehandle
264 my $old_fh = select $fh;
265 my $old_af = $|; $| = 1; $| = $old_af;
268 my @stats = stat($fh);
269 $self->_root->{inode} = $stats[1];
270 $self->_root->{end} = $stats[7];
276 # Check signature was valid
278 unless ($signature eq SIG_FILE) {
280 return $self->_throw_error("Signature not found -- file is not a Deep DB");
283 my @stats = stat($fh);
284 $self->_root->{inode} = $stats[1];
285 $self->_root->{end} = $stats[7];
288 # Get our type from master index signature
290 my $tag = $self->_load_tag($self->_base_offset);
292 #XXX We probably also want to store the hash algorithm name and not assume anything
293 #XXX The cool thing would be to allow a different hashing algorithm at every level
296 return $self->_throw_error("Corrupted file, no master index record");
298 if ($self->{type} ne $tag->{signature}) {
299 return $self->_throw_error("File type mismatch");
309 my $self = $_[0]->_get_self;
310 close $self->_root->{fh} if $self->_root->{fh};
311 $self->_root->{fh} = undef;
316 # Given offset, signature and content, create tag and write to disk
318 my ($self, $offset, $sig, $content) = @_;
319 my $size = length($content);
323 seek($fh, $offset + $self->_root->{file_offset}, SEEK_SET);
324 print( $fh $sig . pack($DATA_LENGTH_PACK, $size) . $content );
326 if ($offset == $self->_root->{end}) {
327 $self->_root->{end} += SIG_SIZE + $DATA_LENGTH_SIZE + $size;
333 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
340 # Given offset, load single tag and return signature, size and data
347 seek($fh, $offset + $self->_root->{file_offset}, SEEK_SET);
348 if (eof $fh) { return undef; }
351 read( $fh, $b, SIG_SIZE + $DATA_LENGTH_SIZE );
352 my ($sig, $size) = unpack( "A $DATA_LENGTH_PACK", $b );
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, $orig_key) = @_;
385 my $keys = $tag->{content};
389 my $root = $self->_root;
391 my $is_dbm_deep = eval { local $SIG{'__DIE__'}; $value->isa( 'DBM::Deep' ) };
392 my $internal_ref = $is_dbm_deep && ($value->_root eq $root);
397 # Iterate through buckets, seeing if this is a new entry or a replace.
399 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
400 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
403 # Found empty bucket (end of list). Populate and exit loop.
407 $location = $internal_ref
408 ? $value->_base_offset
411 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $root->{file_offset}, SEEK_SET);
412 print( $fh $md5 . pack($LONG_PACK, $location) );
416 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
419 # Found existing bucket with same key. Replace with new value.
424 $location = $value->_base_offset;
425 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $root->{file_offset}, SEEK_SET);
426 print( $fh $md5 . pack($LONG_PACK, $location) );
430 seek($fh, $subloc + SIG_SIZE + $root->{file_offset}, SEEK_SET);
432 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
435 # If value is a hash, array, or raw value with equal or less size, we can
436 # reuse the same content area of the database. Otherwise, we have to create
437 # a new content area at the EOF.
440 my $r = Scalar::Util::reftype( $value ) || '';
441 if ( $r eq 'HASH' || $r eq 'ARRAY' ) {
442 $actual_length = $INDEX_SIZE;
444 # if autobless is enabled, must also take into consideration
445 # the class name, as it is stored along with key/value.
446 if ( $root->{autobless} ) {
447 my $value_class = Scalar::Util::blessed($value);
448 if ( defined $value_class && !$value->isa('DBM::Deep') ) {
449 $actual_length += length($value_class);
453 else { $actual_length = length($value); }
455 if ($actual_length <= $size) {
459 $location = $root->{end};
460 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE + $root->{file_offset}, SEEK_SET);
461 print( $fh pack($LONG_PACK, $location) );
469 # If this is an internal reference, return now.
470 # No need to write value or plain key
477 # If bucket didn't fit into list, split into a new index level
480 seek($fh, $tag->{ref_loc} + $root->{file_offset}, SEEK_SET);
481 print( $fh pack($LONG_PACK, $root->{end}) );
483 my $index_tag = $self->_create_tag($root->{end}, SIG_INDEX, chr(0) x $INDEX_SIZE);
486 $keys .= $md5 . pack($LONG_PACK, 0);
488 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
489 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
491 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
492 my $num = ord(substr($key, $tag->{ch} + 1, 1));
494 if ($offsets[$num]) {
495 my $offset = $offsets[$num] + SIG_SIZE + $DATA_LENGTH_SIZE;
496 seek($fh, $offset + $root->{file_offset}, SEEK_SET);
498 read( $fh, $subkeys, $BUCKET_LIST_SIZE);
500 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
501 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
503 seek($fh, $offset + ($k * $BUCKET_SIZE) + $root->{file_offset}, SEEK_SET);
504 print( $fh $key . pack($LONG_PACK, $old_subloc || $root->{end}) );
510 $offsets[$num] = $root->{end};
511 seek($fh, $index_tag->{offset} + ($num * $LONG_SIZE) + $root->{file_offset}, SEEK_SET);
512 print( $fh pack($LONG_PACK, $root->{end}) );
514 my $blist_tag = $self->_create_tag($root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
516 seek($fh, $blist_tag->{offset} + $root->{file_offset}, SEEK_SET);
517 print( $fh $key . pack($LONG_PACK, $old_subloc || $root->{end}) );
522 $location ||= $root->{end};
523 } # re-index bucket list
526 # Seek to content area and store signature, value and plaintext key
530 seek($fh, $location + $root->{file_offset}, SEEK_SET);
533 # Write signature based on content type, set content length and write actual value.
535 my $r = Scalar::Util::reftype($value) || '';
537 print( $fh TYPE_HASH );
538 print( $fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
539 $content_length = $INDEX_SIZE;
541 elsif ($r eq 'ARRAY') {
542 print( $fh TYPE_ARRAY );
543 print( $fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
544 $content_length = $INDEX_SIZE;
546 elsif (!defined($value)) {
547 print( $fh SIG_NULL );
548 print( $fh pack($DATA_LENGTH_PACK, 0) );
552 print( $fh SIG_DATA );
553 print( $fh pack($DATA_LENGTH_PACK, length($value)) . $value );
554 $content_length = length($value);
558 # Plain key is stored AFTER value, as keys are typically fetched less often.
560 print( $fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
563 # If value is blessed, preserve class name
565 if ( $root->{autobless} ) {
566 my $value_class = Scalar::Util::blessed($value);
567 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
569 # Blessed ref -- will restore later
572 print( $fh pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
573 $content_length += 1;
574 $content_length += $DATA_LENGTH_SIZE + length($value_class);
578 $content_length += 1;
583 # If this is a new content area, advance EOF counter
585 if ($location == $root->{end}) {
586 $root->{end} += SIG_SIZE;
587 $root->{end} += $DATA_LENGTH_SIZE + $content_length;
588 $root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
592 # If content is a hash or array, create new child DBM::Deep object and
593 # pass each key or element to it.
596 my $branch = DBM::Deep->new(
598 base_offset => $location,
601 parent_key => $orig_key,
603 foreach my $key (keys %{$value}) {
604 $branch->STORE( $key, $value->{$key} );
607 elsif ($r eq 'ARRAY') {
608 my $branch = DBM::Deep->new(
610 base_offset => $location,
613 parent_key => $orig_key,
616 foreach my $element (@{$value}) {
617 $branch->STORE( $index, $element );
625 return $self->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
628 sub _get_bucket_value {
630 # Fetch single value given tag and MD5 digested key.
633 my ($tag, $md5, $plain_key) = @_;
634 my $keys = $tag->{content};
639 # Iterate through buckets, looking for a key match
642 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
643 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
644 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
648 # Hit end of list, no match
652 if ( $md5 ne $key ) {
657 # Found match -- seek to offset and read signature
660 seek($fh, $subloc + $self->_root->{file_offset}, SEEK_SET);
661 read( $fh, $signature, SIG_SIZE);
664 # If value is a hash or array, return new DBM::Deep object with correct offset
666 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
667 my $obj = DBM::Deep->new(
669 base_offset => $subloc,
670 root => $self->_root,
672 parent_key => $plain_key,
675 if ($self->_root->{autobless}) {
677 # Skip over value and plain key to see if object needs
680 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
683 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
684 if ($size) { seek($fh, $size, SEEK_CUR); }
687 read( $fh, $bless_bit, 1);
688 if (ord($bless_bit)) {
690 # Yes, object needs to be re-blessed
693 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
694 if ($size) { read( $fh, $class_name, $size); }
695 if ($class_name) { $obj = bless( $obj, $class_name ); }
703 # Otherwise return actual value
705 elsif ($signature eq SIG_DATA) {
708 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
709 if ($size) { read( $fh, $value, $size); }
714 # Key exists, but content is null
724 # Delete single key/value pair given tag and MD5 digested key.
727 my ($tag, $md5) = @_;
728 my $keys = $tag->{content};
733 # Iterate through buckets, looking for a key match
736 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
737 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
738 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
742 # Hit end of list, no match
747 if ( $md5 ne $key ) {
752 # Matched key -- delete bucket and return
754 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $self->_root->{file_offset}, SEEK_SET);
755 print( $fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
756 print( $fh chr(0) x $BUCKET_SIZE );
766 # Check existence of single key given tag and MD5 digested key.
769 my ($tag, $md5) = @_;
770 my $keys = $tag->{content};
773 # Iterate through buckets, looking for a key match
776 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
777 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
778 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
782 # Hit end of list, no match
787 if ( $md5 ne $key ) {
792 # Matched key -- return true
800 sub _find_bucket_list {
802 # Locate offset for bucket list, given digested key
808 # Locate offset for bucket list using digest index system
811 my $tag = $self->_load_tag($self->_base_offset);
812 if (!$tag) { return; }
814 while ($tag->{signature} ne SIG_BLIST) {
815 $tag = $self->_index_lookup($tag, ord(substr($md5, $ch, 1)));
816 if (!$tag) { return; }
823 sub _traverse_index {
825 # Scan index and recursively step into deeper levels, looking for next key.
827 my ($self, $offset, $ch, $force_return_next) = @_;
828 $force_return_next = undef unless $force_return_next;
830 my $tag = $self->_load_tag( $offset );
834 if ($tag->{signature} ne SIG_BLIST) {
835 my $content = $tag->{content};
837 if ($self->{return_next}) { $start = 0; }
838 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
840 for (my $index = $start; $index < 256; $index++) {
841 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
843 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
844 if (defined($result)) { return $result; }
848 $self->{return_next} = 1;
851 elsif ($tag->{signature} eq SIG_BLIST) {
852 my $keys = $tag->{content};
853 if ($force_return_next) { $self->{return_next} = 1; }
856 # Iterate through buckets, looking for a key match
858 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
859 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
860 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
864 # End of bucket list -- return to outer loop
866 $self->{return_next} = 1;
869 elsif ($key eq $self->{prev_md5}) {
871 # Located previous key -- return next one found
873 $self->{return_next} = 1;
876 elsif ($self->{return_next}) {
878 # Seek to bucket location and skip over signature
880 seek($fh, $subloc + SIG_SIZE + $self->_root->{file_offset}, SEEK_SET);
883 # Skip over value to get to plain key
886 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
887 if ($size) { seek($fh, $size, SEEK_CUR); }
890 # Read in plain key and return as scalar
893 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
894 if ($size) { read( $fh, $plain_key, $size); }
900 $self->{return_next} = 1;
901 } # tag is a bucket list
908 # Locate next key, given digested previous one
910 my $self = $_[0]->_get_self;
912 $self->{prev_md5} = $_[1] ? $_[1] : undef;
913 $self->{return_next} = 0;
916 # If the previous key was not specifed, start at the top and
917 # return the first one found.
919 if (!$self->{prev_md5}) {
920 $self->{prev_md5} = chr(0) x $HASH_SIZE;
921 $self->{return_next} = 1;
924 return $self->_traverse_index( $self->_base_offset, 0 );
929 # If db locking is set, flock() the db file. If called multiple
930 # times before unlock(), then the same number of unlocks() must
931 # be called before the lock is released.
933 my $self = $_[0]->_get_self;
935 $type = LOCK_EX unless defined $type;
937 if (!defined($self->_fh)) { return; }
939 if ($self->_root->{locking}) {
940 if (!$self->_root->{locked}) {
941 flock($self->_fh, $type);
943 # refresh end counter in case file has changed size
944 my @stats = stat($self->_root->{file});
945 $self->_root->{end} = $stats[7];
947 # double-check file inode, in case another process
948 # has optimize()d our file while we were waiting.
949 if ($stats[1] != $self->_root->{inode}) {
950 $self->_open(); # re-open
951 flock($self->_fh, $type); # re-lock
952 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
955 $self->_root->{locked}++;
965 # If db locking is set, unlock the db file. See note in lock()
966 # regarding calling lock() multiple times.
968 my $self = $_[0]->_get_self;
970 if (!defined($self->_fh)) { return; }
972 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
973 $self->_root->{locked}--;
974 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
983 my $self = shift->_get_self;
984 my ($spot, $value) = @_;
989 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
990 my $type = $value->_type;
991 ${$spot} = $type eq TYPE_HASH ? {} : [];
992 $value->_copy_node( ${$spot} );
995 my $r = Scalar::Util::reftype( $value );
996 my $c = Scalar::Util::blessed( $value );
997 if ( $r eq 'ARRAY' ) {
998 ${$spot} = [ @{$value} ];
1001 ${$spot} = { %{$value} };
1003 ${$spot} = bless ${$spot}, $c
1012 # Copy single level of keys or elements to new DB handle.
1013 # Recurse for nested structures
1015 my $self = shift->_get_self;
1018 if ($self->_type eq TYPE_HASH) {
1019 my $key = $self->first_key();
1021 my $value = $self->get($key);
1022 $self->_copy_value( \$db_temp->{$key}, $value );
1023 $key = $self->next_key($key);
1027 my $length = $self->length();
1028 for (my $index = 0; $index < $length; $index++) {
1029 my $value = $self->get($index);
1030 $self->_copy_value( \$db_temp->[$index], $value );
1039 # Recursively export into standard Perl hashes and arrays.
1041 my $self = $_[0]->_get_self;
1044 if ($self->_type eq TYPE_HASH) { $temp = {}; }
1045 elsif ($self->_type eq TYPE_ARRAY) { $temp = []; }
1048 $self->_copy_node( $temp );
1056 # Recursively import Perl hash/array structure
1058 #XXX This use of ref() seems to be ok
1059 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
1061 my $self = $_[0]->_get_self;
1064 #XXX This use of ref() seems to be ok
1065 if (!ref($struct)) {
1067 # struct is not a reference, so just import based on our type
1071 if ($self->_type eq TYPE_HASH) { $struct = {@_}; }
1072 elsif ($self->_type eq TYPE_ARRAY) { $struct = [@_]; }
1075 my $r = Scalar::Util::reftype($struct) || '';
1076 if ($r eq "HASH" && $self->_type eq TYPE_HASH) {
1077 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
1079 elsif ($r eq "ARRAY" && $self->_type eq TYPE_ARRAY) {
1080 $self->push( @$struct );
1083 return $self->_throw_error("Cannot import: type mismatch");
1091 # Rebuild entire database into new file, then move
1092 # it back on top of original.
1094 my $self = $_[0]->_get_self;
1096 #XXX Need to create a new test for this
1097 # if ($self->_root->{links} > 1) {
1098 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
1101 my $db_temp = DBM::Deep->new(
1102 file => $self->_root->{file} . '.tmp',
1103 type => $self->_type
1106 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
1110 $self->_copy_node( $db_temp );
1114 # Attempt to copy user, group and permissions over to new file
1116 my @stats = stat($self->_fh);
1117 my $perms = $stats[2] & 07777;
1118 my $uid = $stats[4];
1119 my $gid = $stats[5];
1120 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
1121 chmod( $perms, $self->_root->{file} . '.tmp' );
1123 # q.v. perlport for more information on this variable
1124 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
1126 # Potential race condition when optmizing on Win32 with locking.
1127 # The Windows filesystem requires that the filehandle be closed
1128 # before it is overwritten with rename(). This could be redone
1135 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
1136 unlink $self->_root->{file} . '.tmp';
1138 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
1150 # Make copy of object and return
1152 my $self = $_[0]->_get_self;
1154 return DBM::Deep->new(
1155 type => $self->_type,
1156 base_offset => $self->_base_offset,
1157 root => $self->_root,
1158 parent => $self->{parent},
1159 parent_key => $self->{parent_key},
1164 my %is_legal_filter = map {
1167 store_key store_value
1168 fetch_key fetch_value
1173 # Setup filter function for storing or fetching the key or value
1175 my $self = $_[0]->_get_self;
1176 my $type = lc $_[1];
1177 my $func = $_[2] ? $_[2] : undef;
1179 if ( $is_legal_filter{$type} ) {
1180 $self->_root->{"filter_$type"} = $func;
1194 # Get access to the root structure
1196 my $self = $_[0]->_get_self;
1197 return $self->{root};
1202 # Get access to the raw fh
1204 #XXX It will be useful, though, when we split out HASH and ARRAY
1205 my $self = $_[0]->_get_self;
1206 return $self->_root->{fh};
1211 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
1213 my $self = $_[0]->_get_self;
1214 return $self->{type};
1219 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
1221 my $self = $_[0]->_get_self;
1222 return $self->{base_offset};
1227 # Get last error string, or undef if no error
1230 ? ( $_[0]->_get_self->{root}->{error} or undef )
1240 # Store error string in self
1242 my $error_text = $_[1];
1244 if ( Scalar::Util::blessed $_[0] ) {
1245 my $self = $_[0]->_get_self;
1246 $self->_root->{error} = $error_text;
1248 unless ($self->_root->{debug}) {
1249 die "DBM::Deep: $error_text\n";
1252 warn "DBM::Deep: $error_text\n";
1256 die "DBM::Deep: $error_text\n";
1264 my $self = $_[0]->_get_self;
1266 undef $self->_root->{error};
1269 sub _precalc_sizes {
1271 # Precalculate index, bucket and bucket list sizes
1274 #XXX I don't like this ...
1275 set_pack() unless defined $LONG_SIZE;
1277 $INDEX_SIZE = 256 * $LONG_SIZE;
1278 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
1279 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
1284 # Set pack/unpack modes (see file header for more)
1286 my ($long_s, $long_p, $data_s, $data_p) = @_;
1288 $LONG_SIZE = $long_s ? $long_s : 4;
1289 $LONG_PACK = $long_p ? $long_p : 'N';
1291 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
1292 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
1299 # Set key digest function (default is MD5)
1301 my ($digest_func, $hash_size) = @_;
1303 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
1304 $HASH_SIZE = $hash_size ? $hash_size : 16;
1311 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
1316 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
1320 # tie() methods (hashes and arrays)
1325 if ( $self->{parent} ) {
1326 my $base = $self->{parent}->_find_parent();
1327 if ( $self->{parent}->_type eq TYPE_HASH ) {
1328 return $base . "\{$self->{parent_key}\}";
1330 return $base . "\[$self->{parent_key}\]";
1337 # Store single hash key/value or array element in database.
1339 my $self = $_[0]->_get_self;
1342 # User may be storing a hash, in which case we do not want it run
1343 # through the filtering system
1344 my $value = ($self->_root->{filter_store_value} && !ref($_[2]))
1345 ? $self->_root->{filter_store_value}->($_[2])
1348 if ( my $afh = $self->_root->{audit_fh} ) {
1349 unless ( $self->_type eq SIG_ARRAY && $key eq 'length' ) {
1350 my $lhs = $self->_find_parent;
1351 if ( $self->_type eq SIG_HASH ) {
1355 $lhs .= "\[$_[3]\]";
1360 my $r = Scalar::Util::reftype( $_[2] ) || '';
1361 if ( $r eq 'HASH' ) {
1364 elsif ( $r eq 'ARRAY' ) {
1371 if ( my $c = Scalar::Util::blessed( $_[2] ) ) {
1372 $rhs = "bless $rhs, '$c'";
1375 flock( $afh, LOCK_EX );
1376 print( $afh "$lhs = $rhs; # " . localtime(time) . "\n" );
1377 flock( $afh, LOCK_UN );
1381 my $md5 = $DIGEST_FUNC->($key);
1384 # Make sure file is open
1386 if (!defined($self->_fh) && !$self->_open()) {
1390 unless ( _is_writable( $self->_fh ) ) {
1391 $self->_throw_error( 'Cannot write to a readonly filehandle' );
1395 # Request exclusive lock for writing
1397 $self->lock( LOCK_EX );
1399 my $fh = $self->_fh;
1402 # Locate offset for bucket list using digest index system
1404 my $tag = $self->_load_tag($self->_base_offset);
1406 $tag = $self->_create_tag($self->_base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1410 while ($tag->{signature} ne SIG_BLIST) {
1411 my $num = ord(substr($md5, $ch, 1));
1413 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1414 my $new_tag = $self->_index_lookup($tag, $num);
1417 seek($fh, $ref_loc + $self->_root->{file_offset}, SEEK_SET);
1418 print( $fh pack($LONG_PACK, $self->_root->{end}) );
1420 $tag = $self->_create_tag($self->_root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1422 $tag->{ref_loc} = $ref_loc;
1430 $tag->{ref_loc} = $ref_loc;
1437 # Add key/value to bucket list
1439 my $result = $self->_add_bucket( $tag, $md5, $key, $value, $_[3] || $key );
1448 # Fetch single value or element given plain key or array index
1450 my $self = shift->_get_self;
1454 # Make sure file is open
1456 if (!defined($self->_fh)) { $self->_open(); }
1458 my $md5 = $DIGEST_FUNC->($key);
1461 # Request shared lock for reading
1463 $self->lock( LOCK_SH );
1465 my $tag = $self->_find_bucket_list( $md5 );
1472 # Get value from bucket list
1474 my $result = $self->_get_bucket_value( $tag, $md5, $key );
1478 #XXX What is ref() checking here?
1479 #YYY Filters only apply on scalar values, so the ref check is making
1480 #YYY sure the fetched bucket is a scalar, not a child hash or array.
1481 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
1482 ? $self->_root->{filter_fetch_value}->($result)
1488 # Delete single key/value pair or element given plain key or array index
1490 my $self = shift->_get_self;
1491 my ($key, $orig_key) = @_;
1493 if ( my $afh = $self->_root->{audit_fh} ) {
1494 unless ( $self->_type eq SIG_ARRAY && $key eq 'length' ) {
1495 my $lhs = $self->_find_parent;
1496 if ( $self->_type eq SIG_HASH ) {
1500 $lhs .= "\[$_[3]\]";
1503 flock( $afh, LOCK_EX );
1504 print( $afh "delete $lhs; # " . localtime(time) . "\n" );
1505 flock( $afh, LOCK_UN );
1509 my $md5 = $DIGEST_FUNC->($key);
1512 # Make sure file is open
1514 if (!defined($self->_fh)) { $self->_open(); }
1517 # Request exclusive lock for writing
1519 $self->lock( LOCK_EX );
1521 my $tag = $self->_find_bucket_list( $md5 );
1530 my $value = $self->_get_bucket_value( $tag, $md5, $key );
1531 if ($value && !ref($value) && $self->_root->{filter_fetch_value}) {
1532 $value = $self->_root->{filter_fetch_value}->($value);
1535 my $result = $self->_delete_bucket( $tag, $md5 );
1538 # If this object is an array and the key deleted was on the end of the stack,
1539 # decrement the length variable.
1549 # Check if a single key or element exists given plain key or array index
1551 my $self = $_[0]->_get_self;
1554 my $md5 = $DIGEST_FUNC->($key);
1557 # Make sure file is open
1559 if (!defined($self->_fh)) { $self->_open(); }
1562 # Request shared lock for reading
1564 $self->lock( LOCK_SH );
1566 my $tag = $self->_find_bucket_list( $md5 );
1569 # For some reason, the built-in exists() function returns '' for false
1577 # Check if bucket exists and return 1 or ''
1579 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1588 # Clear all keys from hash, or all elements from array.
1590 my $self = $_[0]->_get_self;
1592 if ( my $afh = $self->_root->{audit_fh} ) {
1593 my $lhs = $self->_find_parent;
1596 if ( $self->_type eq SIG_HASH ) {
1599 elsif ( $self->_type eq SIG_ARRAY ) {
1603 flock( $afh, LOCK_EX );
1604 print( $afh "$lhs = $rhs; # " . localtime(time) . "\n" );
1605 flock( $afh, LOCK_UN );
1609 # Make sure file is open
1611 if (!defined($self->_fh)) { $self->_open(); }
1614 # Request exclusive lock for writing
1616 $self->lock( LOCK_EX );
1618 my $fh = $self->_fh;
1620 seek($fh, $self->_base_offset + $self->_root->{file_offset}, SEEK_SET);
1626 $self->_create_tag($self->_base_offset, $self->_type, chr(0) x $INDEX_SIZE);
1634 # Public method aliases
1636 sub put { (shift)->STORE( @_ ) }
1637 sub store { (shift)->STORE( @_ ) }
1638 sub get { (shift)->FETCH( @_ ) }
1639 sub fetch { (shift)->FETCH( @_ ) }
1640 sub delete { (shift)->DELETE( @_ ) }
1641 sub exists { (shift)->EXISTS( @_ ) }
1642 sub clear { (shift)->CLEAR( @_ ) }
1644 package DBM::Deep::_::Root;
1660 filter_store_key => undef,
1661 filter_store_value => undef,
1662 filter_fetch_key => undef,
1663 filter_fetch_value => undef,
1669 if ( $self->{fh} && !$self->{file_offset} ) {
1670 $self->{file_offset} = tell( $self->{fh} );
1673 if ( $self->{audit_file} && !$self->{audit_fh} ) {
1674 my $flags = O_WRONLY | O_APPEND | O_CREAT;
1677 sysopen( $fh, $self->{audit_file}, $flags )
1678 or die "Cannot open audit file: $!";
1680 my $old = select $fh;
1684 $self->{audit_fh} = $fh;
1692 return unless $self;
1694 close $self->{fh} if $self->{fh};
1705 DBM::Deep - A pure perl multi-level hash/array DBM
1710 my $db = DBM::Deep->new( "foo.db" );
1712 $db->{key} = 'value'; # tie() style
1715 $db->put('key' => 'value'); # OO style
1716 print $db->get('key');
1718 # true multi-level support
1719 $db->{my_complex} = [
1720 'hello', { perl => 'rules' },
1726 A unique flat-file database module, written in pure perl. True
1727 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1728 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1729 handle millions of keys and unlimited hash levels without significant
1730 slow-down. Written from the ground-up in pure perl -- this is NOT a
1731 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1732 Mac OS X and Windows.
1736 Hopefully you are using Perl's excellent CPAN module, which will download
1737 and install the module for you. If not, get the tarball, and run these
1749 Construction can be done OO-style (which is the recommended way), or using
1750 Perl's tie() function. Both are examined here.
1752 =head2 OO CONSTRUCTION
1754 The recommended way to construct a DBM::Deep object is to use the new()
1755 method, which gets you a blessed, tied hash or array reference.
1757 my $db = DBM::Deep->new( "foo.db" );
1759 This opens a new database handle, mapped to the file "foo.db". If this
1760 file does not exist, it will automatically be created. DB files are
1761 opened in "r+" (read/write) mode, and the type of object returned is a
1762 hash, unless otherwise specified (see L<OPTIONS> below).
1764 You can pass a number of options to the constructor to specify things like
1765 locking, autoflush, etc. This is done by passing an inline hash:
1767 my $db = DBM::Deep->new(
1773 Notice that the filename is now specified I<inside> the hash with
1774 the "file" parameter, as opposed to being the sole argument to the
1775 constructor. This is required if any options are specified.
1776 See L<OPTIONS> below for the complete list.
1780 You can also start with an array instead of a hash. For this, you must
1781 specify the C<type> parameter:
1783 my $db = DBM::Deep->new(
1785 type => DBM::Deep->TYPE_ARRAY
1788 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1789 a new DB file. If you create a DBM::Deep object with an existing file, the
1790 C<type> will be loaded from the file header, and an error will be thrown if
1791 the wrong type is passed in.
1793 =head2 TIE CONSTRUCTION
1795 Alternately, you can create a DBM::Deep handle by using Perl's built-in
1796 tie() function. The object returned from tie() can be used to call methods,
1797 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
1798 file (as expected with most tie'd objects).
1801 my $db = tie %hash, "DBM::Deep", "foo.db";
1804 my $db = tie @array, "DBM::Deep", "bar.db";
1806 As with the OO constructor, you can replace the DB filename parameter with
1807 a hash containing one or more options (see L<OPTIONS> just below for the
1810 tie %hash, "DBM::Deep", {
1818 There are a number of options that can be passed in when constructing your
1819 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1825 Filename of the DB file to link the handle to. You can pass a full absolute
1826 filesystem path, partial path, or a plain filename if the file is in the
1827 current working directory. This is a required parameter (though q.v. fh).
1831 If you want, you can pass in the fh instead of the file. This is most useful for doing
1834 my $db = DBM::Deep->new( { fh => \*DATA } );
1836 You are responsible for making sure that the fh has been opened appropriately for your
1837 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
1838 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
1839 needs to read from the fh.
1843 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
1844 not need to set this. However, it's there if you want it.
1846 If you pass in fh and do not set this, it will be set appropriately.
1850 This parameter specifies what type of object to create, a hash or array. Use
1851 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1852 This only takes effect when beginning a new file. This is an optional
1853 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1857 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1858 function to lock the database in exclusive mode for writes, and shared mode for
1859 reads. Pass any true value to enable. This affects the base DB handle I<and
1860 any child hashes or arrays> that use the same DB file. This is an optional
1861 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1865 Specifies whether autoflush is to be enabled on the underlying filehandle.
1866 This obviously slows down write operations, but is required if you may have
1867 multiple processes accessing the same DB file (also consider enable I<locking>).
1868 Pass any true value to enable. This is an optional parameter, and defaults to 0
1873 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1874 restore them when fetched. This is an B<experimental> feature, and does have
1875 side-effects. Basically, when hashes are re-blessed into their original
1876 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1877 able to call any DBM::Deep methods on them. You have been warned.
1878 This is an optional parameter, and defaults to 0 (disabled).
1882 See L<FILTERS> below.
1886 Setting I<debug> mode will make all errors non-fatal, dump them out to
1887 STDERR, and continue on. This is for debugging purposes only, and probably
1888 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1890 B<NOTE>: This parameter is considered deprecated and should not be used anymore.
1892 =item * audit_file / audit_fh
1894 If you set either of these, an auditlog will be written to. If you set
1895 audit_file, audit_fh will be set to the open() on the audit_file.
1897 The auditing information will look something like:
1899 $db->{foo} = 'floober';
1902 $db->{bar}{a}[0] = '5';
1904 The idea is that if your DB file is corrupted, you can recover it by doing
1907 my $db = DBM::Deep->new( $new_filename );
1910 It is your responsability to make sure that the same auditlog is opened with the
1911 same DB file every time the DB file is opened. This will change when 1.00 is
1916 =head1 TIE INTERFACE
1918 With DBM::Deep you can access your databases using Perl's standard hash/array
1919 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
1920 treat them as such. DBM::Deep will intercept all reads/writes and direct them
1921 to the right place -- the DB file. This has nothing to do with the
1922 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
1923 using regular hashes and arrays, rather than calling functions like C<get()>
1924 and C<put()> (although those work too). It is entirely up to you how to want
1925 to access your databases.
1929 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1930 or even nested hashes (or arrays) using standard Perl syntax:
1932 my $db = DBM::Deep->new( "foo.db" );
1934 $db->{mykey} = "myvalue";
1936 $db->{myhash}->{subkey} = "subvalue";
1938 print $db->{myhash}->{subkey} . "\n";
1940 You can even step through hash keys using the normal Perl C<keys()> function:
1942 foreach my $key (keys %$db) {
1943 print "$key: " . $db->{$key} . "\n";
1946 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1947 pushes them onto an array, all before the loop even begins. If you have an
1948 extra large hash, this may exhaust Perl's memory. Instead, consider using
1949 Perl's C<each()> function, which pulls keys/values one at a time, using very
1952 while (my ($key, $value) = each %$db) {
1953 print "$key: $value\n";
1956 Please note that when using C<each()>, you should always pass a direct
1957 hash reference, not a lookup. Meaning, you should B<never> do this:
1960 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1962 This causes an infinite loop, because for each iteration, Perl is calling
1963 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1964 it effectively keeps returning the first key over and over again. Instead,
1965 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1969 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1970 reference. This includes inserting, removing and manipulating elements,
1971 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1972 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1973 or simply be a nested array reference inside a hash. Example:
1975 my $db = DBM::Deep->new(
1976 file => "foo-array.db",
1977 type => DBM::Deep->TYPE_ARRAY
1981 push @$db, "bar", "baz";
1982 unshift @$db, "bah";
1984 my $last_elem = pop @$db; # baz
1985 my $first_elem = shift @$db; # bah
1986 my $second_elem = $db->[1]; # bar
1988 my $num_elements = scalar @$db;
1992 In addition to the I<tie()> interface, you can also use a standard OO interface
1993 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1994 array) has its own methods, but both types share the following common methods:
1995 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1999 =item * new() / clone()
2001 These are the constructor and copy-functions.
2003 =item * put() / store()
2005 Stores a new hash key/value pair, or sets an array element value. Takes two
2006 arguments, the hash key or array index, and the new value. The value can be
2007 a scalar, hash ref or array ref. Returns true on success, false on failure.
2009 $db->put("foo", "bar"); # for hashes
2010 $db->put(1, "bar"); # for arrays
2012 =item * get() / fetch()
2014 Fetches the value of a hash key or array element. Takes one argument: the hash
2015 key or array index. Returns a scalar, hash ref or array ref, depending on the
2018 my $value = $db->get("foo"); # for hashes
2019 my $value = $db->get(1); # for arrays
2023 Checks if a hash key or array index exists. Takes one argument: the hash key
2024 or array index. Returns true if it exists, false if not.
2026 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
2027 if ($db->exists(1)) { print "yay!\n"; } # for arrays
2031 Deletes one hash key/value pair or array element. Takes one argument: the hash
2032 key or array index. Returns true on success, false if not found. For arrays,
2033 the remaining elements located after the deleted element are NOT moved over.
2034 The deleted element is essentially just undefined, which is exactly how Perl's
2035 internal arrays work. Please note that the space occupied by the deleted
2036 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
2037 below for details and workarounds.
2039 $db->delete("foo"); # for hashes
2040 $db->delete(1); # for arrays
2044 Deletes B<all> hash keys or array elements. Takes no arguments. No return
2045 value. Please note that the space occupied by the deleted keys/values or
2046 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
2047 details and workarounds.
2049 $db->clear(); # hashes or arrays
2051 =item * lock() / unlock()
2057 Recover lost disk space.
2059 =item * import() / export()
2061 Data going in and out.
2063 =item * set_digest() / set_pack() / set_filter()
2065 q.v. adjusting the interal parameters.
2067 =item * error() / clear_error()
2069 Error handling methods. These are deprecated and will be removed in 1.00.
2075 For hashes, DBM::Deep supports all the common methods described above, and the
2076 following additional methods: C<first_key()> and C<next_key()>.
2082 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
2083 fetched in an undefined order (which appears random). Takes no arguments,
2084 returns the key as a scalar value.
2086 my $key = $db->first_key();
2090 Returns the "next" key in the hash, given the previous one as the sole argument.
2091 Returns undef if there are no more keys to be fetched.
2093 $key = $db->next_key($key);
2097 Here are some examples of using hashes:
2099 my $db = DBM::Deep->new( "foo.db" );
2101 $db->put("foo", "bar");
2102 print "foo: " . $db->get("foo") . "\n";
2104 $db->put("baz", {}); # new child hash ref
2105 $db->get("baz")->put("buz", "biz");
2106 print "buz: " . $db->get("baz")->get("buz") . "\n";
2108 my $key = $db->first_key();
2110 print "$key: " . $db->get($key) . "\n";
2111 $key = $db->next_key($key);
2114 if ($db->exists("foo")) { $db->delete("foo"); }
2118 For arrays, DBM::Deep supports all the common methods described above, and the
2119 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
2120 C<unshift()> and C<splice()>.
2126 Returns the number of elements in the array. Takes no arguments.
2128 my $len = $db->length();
2132 Adds one or more elements onto the end of the array. Accepts scalars, hash
2133 refs or array refs. No return value.
2135 $db->push("foo", "bar", {});
2139 Fetches the last element in the array, and deletes it. Takes no arguments.
2140 Returns undef if array is empty. Returns the element value.
2142 my $elem = $db->pop();
2146 Fetches the first element in the array, deletes it, then shifts all the
2147 remaining elements over to take up the space. Returns the element value. This
2148 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
2151 my $elem = $db->shift();
2155 Inserts one or more elements onto the beginning of the array, shifting all
2156 existing elements over to make room. Accepts scalars, hash refs or array refs.
2157 No return value. This method is not recommended with large arrays -- see
2158 <LARGE ARRAYS> below for details.
2160 $db->unshift("foo", "bar", {});
2164 Performs exactly like Perl's built-in function of the same name. See L<perldoc
2165 -f splice> for usage -- it is too complicated to document here. This method is
2166 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
2170 Here are some examples of using arrays:
2172 my $db = DBM::Deep->new(
2174 type => DBM::Deep->TYPE_ARRAY
2177 $db->push("bar", "baz");
2178 $db->unshift("foo");
2181 my $len = $db->length();
2182 print "length: $len\n"; # 4
2184 for (my $k=0; $k<$len; $k++) {
2185 print "$k: " . $db->get($k) . "\n";
2188 $db->splice(1, 2, "biz", "baf");
2190 while (my $elem = shift @$db) {
2191 print "shifted: $elem\n";
2196 Enable automatic file locking by passing a true value to the C<locking>
2197 parameter when constructing your DBM::Deep object (see L<SETUP> above).
2199 my $db = DBM::Deep->new(
2204 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
2205 mode for writes, and shared mode for reads. This is required if you have
2206 multiple processes accessing the same database file, to avoid file corruption.
2207 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
2208 NFS> below for more.
2210 =head2 EXPLICIT LOCKING
2212 You can explicitly lock a database, so it remains locked for multiple
2213 transactions. This is done by calling the C<lock()> method, and passing an
2214 optional lock mode argument (defaults to exclusive mode). This is particularly
2215 useful for things like counters, where the current value needs to be fetched,
2216 then incremented, then stored again.
2219 my $counter = $db->get("counter");
2221 $db->put("counter", $counter);
2230 You can pass C<lock()> an optional argument, which specifies which mode to use
2231 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2232 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2233 same as the constants defined in Perl's C<Fcntl> module.
2235 $db->lock( DBM::Deep->LOCK_SH );
2239 =head1 IMPORTING/EXPORTING
2241 You can import existing complex structures by calling the C<import()> method,
2242 and export an entire database into an in-memory structure using the C<export()>
2243 method. Both are examined here.
2247 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2248 walking the structure and adding keys/elements to the database as you go,
2249 simply pass a reference to the C<import()> method. This recursively adds
2250 everything to an existing DBM::Deep object for you. Here is an example:
2255 array1 => [ "elem0", "elem1", "elem2" ],
2257 subkey1 => "subvalue1",
2258 subkey2 => "subvalue2"
2262 my $db = DBM::Deep->new( "foo.db" );
2263 $db->import( $struct );
2265 print $db->{key1} . "\n"; # prints "value1"
2267 This recursively imports the entire C<$struct> object into C<$db>, including
2268 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2269 keys are merged with the existing ones, replacing if they already exist.
2270 The C<import()> method can be called on any database level (not just the base
2271 level), and works with both hash and array DB types.
2273 B<Note:> Make sure your existing structure has no circular references in it.
2274 These will cause an infinite loop when importing.
2278 Calling the C<export()> method on an existing DBM::Deep object will return
2279 a reference to a new in-memory copy of the database. The export is done
2280 recursively, so all nested hashes/arrays are all exported to standard Perl
2281 objects. Here is an example:
2283 my $db = DBM::Deep->new( "foo.db" );
2285 $db->{key1} = "value1";
2286 $db->{key2} = "value2";
2288 $db->{hash1}->{subkey1} = "subvalue1";
2289 $db->{hash1}->{subkey2} = "subvalue2";
2291 my $struct = $db->export();
2293 print $struct->{key1} . "\n"; # prints "value1"
2295 This makes a complete copy of the database in memory, and returns a reference
2296 to it. The C<export()> method can be called on any database level (not just
2297 the base level), and works with both hash and array DB types. Be careful of
2298 large databases -- you can store a lot more data in a DBM::Deep object than an
2299 in-memory Perl structure.
2301 B<Note:> Make sure your database has no circular references in it.
2302 These will cause an infinite loop when exporting.
2306 DBM::Deep has a number of hooks where you can specify your own Perl function
2307 to perform filtering on incoming or outgoing data. This is a perfect
2308 way to extend the engine, and implement things like real-time compression or
2309 encryption. Filtering applies to the base DB level, and all child hashes /
2310 arrays. Filter hooks can be specified when your DBM::Deep object is first
2311 constructed, or by calling the C<set_filter()> method at any time. There are
2312 four available filter hooks, described below:
2316 =item * filter_store_key
2318 This filter is called whenever a hash key is stored. It
2319 is passed the incoming key, and expected to return a transformed key.
2321 =item * filter_store_value
2323 This filter is called whenever a hash key or array element is stored. It
2324 is passed the incoming value, and expected to return a transformed value.
2326 =item * filter_fetch_key
2328 This filter is called whenever a hash key is fetched (i.e. via
2329 C<first_key()> or C<next_key()>). It is passed the transformed key,
2330 and expected to return the plain key.
2332 =item * filter_fetch_value
2334 This filter is called whenever a hash key or array element is fetched.
2335 It is passed the transformed value, and expected to return the plain value.
2339 Here are the two ways to setup a filter hook:
2341 my $db = DBM::Deep->new(
2343 filter_store_value => \&my_filter_store,
2344 filter_fetch_value => \&my_filter_fetch
2349 $db->set_filter( "filter_store_value", \&my_filter_store );
2350 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2352 Your filter function will be called only when dealing with SCALAR keys or
2353 values. When nested hashes and arrays are being stored/fetched, filtering
2354 is bypassed. Filters are called as static functions, passed a single SCALAR
2355 argument, and expected to return a single SCALAR value. If you want to
2356 remove a filter, set the function reference to C<undef>:
2358 $db->set_filter( "filter_store_value", undef );
2360 =head2 REAL-TIME ENCRYPTION EXAMPLE
2362 Here is a working example that uses the I<Crypt::Blowfish> module to
2363 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2364 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2365 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2368 use Crypt::Blowfish;
2371 my $cipher = Crypt::CBC->new({
2372 'key' => 'my secret key',
2373 'cipher' => 'Blowfish',
2375 'regenerate_key' => 0,
2376 'padding' => 'space',
2380 my $db = DBM::Deep->new(
2381 file => "foo-encrypt.db",
2382 filter_store_key => \&my_encrypt,
2383 filter_store_value => \&my_encrypt,
2384 filter_fetch_key => \&my_decrypt,
2385 filter_fetch_value => \&my_decrypt,
2388 $db->{key1} = "value1";
2389 $db->{key2} = "value2";
2390 print "key1: " . $db->{key1} . "\n";
2391 print "key2: " . $db->{key2} . "\n";
2397 return $cipher->encrypt( $_[0] );
2400 return $cipher->decrypt( $_[0] );
2403 =head2 REAL-TIME COMPRESSION EXAMPLE
2405 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2406 compression / decompression of keys & values with DBM::Deep Filters.
2407 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2408 more on I<Compress::Zlib>.
2413 my $db = DBM::Deep->new(
2414 file => "foo-compress.db",
2415 filter_store_key => \&my_compress,
2416 filter_store_value => \&my_compress,
2417 filter_fetch_key => \&my_decompress,
2418 filter_fetch_value => \&my_decompress,
2421 $db->{key1} = "value1";
2422 $db->{key2} = "value2";
2423 print "key1: " . $db->{key1} . "\n";
2424 print "key2: " . $db->{key2} . "\n";
2430 return Compress::Zlib::memGzip( $_[0] ) ;
2433 return Compress::Zlib::memGunzip( $_[0] ) ;
2436 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2437 actually numerical index numbers, and are not filtered.
2439 =head1 ERROR HANDLING
2441 Most DBM::Deep methods return a true value for success, and call die() on
2442 failure. You can wrap calls in an eval block to catch the die. Also, the
2443 actual error message is stored in an internal scalar, which can be fetched by
2444 calling the C<error()> method.
2446 my $db = DBM::Deep->new( "foo.db" ); # create hash
2447 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2449 print $@; # prints error message
2450 print $db->error(); # prints error message
2452 You can then call C<clear_error()> to clear the current error state.
2456 If you set the C<debug> option to true when creating your DBM::Deep object,
2457 all errors are considered NON-FATAL, and dumped to STDERR. This should only
2458 be used for debugging purposes and not production work. DBM::Deep expects errors
2459 to be thrown, not propagated back up the stack.
2461 B<NOTE>: error() and clear_error() are considered deprecated and I<will> be removed
2462 in 1.00. Please don't use them. Instead, wrap all your functions with in eval-blocks.
2464 =head1 LARGEFILE SUPPORT
2466 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2467 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2468 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2469 by calling the static C<set_pack()> method before you do anything else.
2471 DBM::Deep::set_pack(8, 'Q');
2473 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2474 instead of 32-bit longs. After setting these values your DB files have a
2475 theoretical maximum size of 16 XB (exabytes).
2477 B<Note:> Changing these values will B<NOT> work for existing database files.
2478 Only change this for new files, and make sure it stays set consistently
2479 throughout the file's life. If you do set these values, you can no longer
2480 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2481 back to 32-bit mode.
2483 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2484 only a 32-bit Perl. However, I have received user reports that this does
2487 =head1 LOW-LEVEL ACCESS
2489 If you require low-level access to the underlying filehandle that DBM::Deep uses,
2490 you can call the C<_fh()> method, which returns the handle:
2492 my $fh = $db->_fh();
2494 This method can be called on the root level of the datbase, or any child
2495 hashes or arrays. All levels share a I<root> structure, which contains things
2496 like the filehandle, a reference counter, and all the options specified
2497 when you created the object. You can get access to this root structure by
2498 calling the C<root()> method.
2500 my $root = $db->_root();
2502 This is useful for changing options after the object has already been created,
2503 such as enabling/disabling locking, or debug modes. You can also
2504 store your own temporary user data in this structure (be wary of name
2505 collision), which is then accessible from any child hash or array.
2507 =head1 CUSTOM DIGEST ALGORITHM
2509 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2510 keys. However you can override this, and use another algorithm (such as SHA-256)
2511 or even write your own. But please note that DBM::Deep currently expects zero
2512 collisions, so your algorithm has to be I<perfect>, so to speak.
2513 Collision detection may be introduced in a later version.
2517 You can specify a custom digest algorithm by calling the static C<set_digest()>
2518 function, passing a reference to a subroutine, and the length of the algorithm's
2519 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2520 objects. Here is a working example that uses a 256-bit hash from the
2521 I<Digest::SHA256> module. Please see
2522 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2527 my $context = Digest::SHA256::new(256);
2529 DBM::Deep::set_digest( \&my_digest, 32 );
2531 my $db = DBM::Deep->new( "foo-sha.db" );
2533 $db->{key1} = "value1";
2534 $db->{key2} = "value2";
2535 print "key1: " . $db->{key1} . "\n";
2536 print "key2: " . $db->{key2} . "\n";
2542 return substr( $context->hash($_[0]), 0, 32 );
2545 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2546 of bytes you specify in the C<set_digest()> function (in this case 32).
2548 =head1 CIRCULAR REFERENCES
2550 DBM::Deep has B<experimental> support for circular references. Meaning you
2551 can have a nested hash key or array element that points to a parent object.
2552 This relationship is stored in the DB file, and is preserved between sessions.
2555 my $db = DBM::Deep->new( "foo.db" );
2558 $db->{circle} = $db; # ref to self
2560 print $db->{foo} . "\n"; # prints "foo"
2561 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2563 One catch is, passing the object to a function that recursively walks the
2564 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2565 C<export()> methods) will result in an infinite loop. The other catch is,
2566 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2567 or C<next_key()> methods), you will get the I<target object's key>, not the
2568 ref's key. This gets even more interesting with the above example, where
2569 the I<circle> key points to the base DB object, which technically doesn't
2570 have a key. So I made DBM::Deep return "[base]" as the key name in that
2573 =head1 CAVEATS / ISSUES / BUGS
2575 This section describes all the known issues with DBM::Deep. It you have found
2576 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2578 =head2 UNUSED SPACE RECOVERY
2580 One major caveat with DBM::Deep is that space occupied by existing keys and
2581 values is not recovered when they are deleted. Meaning if you keep deleting
2582 and adding new keys, your file will continuously grow. I am working on this,
2583 but in the meantime you can call the built-in C<optimize()> method from time to
2584 time (perhaps in a crontab or something) to recover all your unused space.
2586 $db->optimize(); # returns true on success
2588 This rebuilds the ENTIRE database into a new file, then moves it on top of
2589 the original. The new file will have no unused space, thus it will take up as
2590 little disk space as possible. Please note that this operation can take
2591 a long time for large files, and you need enough disk space to temporarily hold
2592 2 copies of your DB file. The temporary file is created in the same directory
2593 as the original, named with a ".tmp" extension, and is deleted when the
2594 operation completes. Oh, and if locking is enabled, the DB is automatically
2595 locked for the entire duration of the copy.
2597 B<WARNING:> Only call optimize() on the top-level node of the database, and
2598 make sure there are no child references lying around. DBM::Deep keeps a reference
2599 counter, and if it is greater than 1, optimize() will abort and return undef.
2601 =head2 AUTOVIVIFICATION
2603 Unfortunately, autovivification doesn't work with tied hashes. This appears to
2604 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
2605 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
2606 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
2607 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
2610 $db->{foo}->{bar} = "hello";
2612 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
2613 being an empty hash. Try this instead, which works fine:
2615 $db->{foo} = { bar => "hello" };
2617 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
2618 the execution path, and Perl indeed passes an empty hash to the STORE() method.
2619 Probably a bug in Perl.
2621 =head2 FILE CORRUPTION
2623 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2624 for a 32-bit signature when opened, but other corruption in files can cause
2625 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2626 stuck in an infinite loop depending on the level of corruption. File write
2627 operations are not checked for failure (for speed), so if you happen to run
2628 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2629 be addressed in a later version of DBM::Deep.
2633 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2634 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2635 about setting up your NFS server with a locking daemon, then using lockf() to
2636 lock your files, but your mileage may vary there as well. From what I
2637 understand, there is no real way to do it. However, if you need access to the
2638 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
2639 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2641 =head2 COPYING OBJECTS
2643 Beware of copying tied objects in Perl. Very strange things can happen.
2644 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2645 returns a new, blessed, tied hash or array to the same level in the DB.
2647 my $copy = $db->clone();
2649 B<Note>: Since clone() here is cloning the object, not the database location, any
2650 modifications to either $db or $copy will be visible in both.
2654 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2655 These functions cause every element in the array to move, which can be murder
2656 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2657 a different location. This will be addressed in the forthcoming version 1.00.
2659 =head2 WRITEONLY FILES
2661 If you pass in a filehandle to new(), you may have opened it in either a readonly or
2662 writeonly mode. STORE will verify that the filehandle is writable. However, there
2663 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
2664 filehandle isn't readable, it's not clear what will happen. So, don't do that.
2668 This section discusses DBM::Deep's speed and memory usage.
2672 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2673 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2674 multi-level hash/array support, and cross-platform FTPable files. Even so,
2675 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2676 with huge databases. Here is some test data:
2678 Adding 1,000,000 keys to new DB file...
2680 At 100 keys, avg. speed is 2,703 keys/sec
2681 At 200 keys, avg. speed is 2,642 keys/sec
2682 At 300 keys, avg. speed is 2,598 keys/sec
2683 At 400 keys, avg. speed is 2,578 keys/sec
2684 At 500 keys, avg. speed is 2,722 keys/sec
2685 At 600 keys, avg. speed is 2,628 keys/sec
2686 At 700 keys, avg. speed is 2,700 keys/sec
2687 At 800 keys, avg. speed is 2,607 keys/sec
2688 At 900 keys, avg. speed is 2,190 keys/sec
2689 At 1,000 keys, avg. speed is 2,570 keys/sec
2690 At 2,000 keys, avg. speed is 2,417 keys/sec
2691 At 3,000 keys, avg. speed is 1,982 keys/sec
2692 At 4,000 keys, avg. speed is 1,568 keys/sec
2693 At 5,000 keys, avg. speed is 1,533 keys/sec
2694 At 6,000 keys, avg. speed is 1,787 keys/sec
2695 At 7,000 keys, avg. speed is 1,977 keys/sec
2696 At 8,000 keys, avg. speed is 2,028 keys/sec
2697 At 9,000 keys, avg. speed is 2,077 keys/sec
2698 At 10,000 keys, avg. speed is 2,031 keys/sec
2699 At 20,000 keys, avg. speed is 1,970 keys/sec
2700 At 30,000 keys, avg. speed is 2,050 keys/sec
2701 At 40,000 keys, avg. speed is 2,073 keys/sec
2702 At 50,000 keys, avg. speed is 1,973 keys/sec
2703 At 60,000 keys, avg. speed is 1,914 keys/sec
2704 At 70,000 keys, avg. speed is 2,091 keys/sec
2705 At 80,000 keys, avg. speed is 2,103 keys/sec
2706 At 90,000 keys, avg. speed is 1,886 keys/sec
2707 At 100,000 keys, avg. speed is 1,970 keys/sec
2708 At 200,000 keys, avg. speed is 2,053 keys/sec
2709 At 300,000 keys, avg. speed is 1,697 keys/sec
2710 At 400,000 keys, avg. speed is 1,838 keys/sec
2711 At 500,000 keys, avg. speed is 1,941 keys/sec
2712 At 600,000 keys, avg. speed is 1,930 keys/sec
2713 At 700,000 keys, avg. speed is 1,735 keys/sec
2714 At 800,000 keys, avg. speed is 1,795 keys/sec
2715 At 900,000 keys, avg. speed is 1,221 keys/sec
2716 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2718 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2719 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2720 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2721 Run time was 12 min 3 sec.
2725 One of the great things about DBM::Deep is that it uses very little memory.
2726 Even with huge databases (1,000,000+ keys) you will not see much increased
2727 memory on your process. DBM::Deep relies solely on the filesystem for storing
2728 and fetching data. Here is output from I</usr/bin/top> before even opening a
2731 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2732 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2734 Basically the process is taking 2,716K of memory. And here is the same
2735 process after storing and fetching 1,000,000 keys:
2737 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2738 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2740 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2741 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2743 =head1 DB FILE FORMAT
2745 In case you were interested in the underlying DB file format, it is documented
2746 here in this section. You don't need to know this to use the module, it's just
2747 included for reference.
2751 DBM::Deep files always start with a 32-bit signature to identify the file type.
2752 This is at offset 0. The signature is "DPDB" in network byte order. This is
2753 checked for when the file is opened and an error will be thrown if it's not found.
2757 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2758 has a standard header containing the type of data, the length of data, and then
2759 the data itself. The type is a single character (1 byte), the length is a
2760 32-bit unsigned long in network byte order, and the data is, well, the data.
2761 Here is how it unfolds:
2765 Immediately after the 32-bit file signature is the I<Master Index> record.
2766 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2767 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2768 depending on how the DBM::Deep object was constructed.
2770 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2771 number). The first 8-bit char of the MD5 signature is the offset into the
2772 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2773 index element is a file offset of the next tag for the key/element in question,
2774 which is usually a I<Bucket List> tag (see below).
2776 The next tag I<could> be another index, depending on how many keys/elements
2777 exist. See L<RE-INDEXING> below for details.
2781 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2782 file offsets to where the actual data is stored. It starts with a standard
2783 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2784 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2785 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2786 When the list fills up, a I<Re-Index> operation is performed (See
2787 L<RE-INDEXING> below).
2791 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2792 index/value pair (in array mode). It starts with a standard tag header with
2793 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2794 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2795 header. The size reported in the tag header is only for the value, but then,
2796 just after the value is another size (32-bit unsigned long) and then the plain
2797 key itself. Since the value is likely to be fetched more often than the plain
2798 key, I figured it would be I<slightly> faster to store the value first.
2800 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2801 record for the nested structure, where the process begins all over again.
2805 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2806 exhausted. Then, when another key/element comes in, the list is converted to a
2807 new index record. However, this index will look at the next char in the MD5
2808 hash, and arrange new Bucket List pointers accordingly. This process is called
2809 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2810 17 (16 + new one) keys/elements are removed from the old Bucket List and
2811 inserted into the new index. Several new Bucket Lists are created in the
2812 process, as a new MD5 char from the key is being examined (it is unlikely that
2813 the keys will all share the same next char of their MD5s).
2815 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2816 when the Bucket Lists will turn into indexes, but the first round tends to
2817 happen right around 4,000 keys. You will see a I<slight> decrease in
2818 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2819 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2820 right around 900,000 keys. This process can continue nearly indefinitely --
2821 right up until the point the I<MD5> signatures start colliding with each other,
2822 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2823 getting struck by lightning while you are walking to cash in your tickets.
2824 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2825 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2826 this is 340 unodecillion, but don't quote me).
2830 When a new key/element is stored, the key (or index number) is first run through
2831 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2832 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2833 for the first char of the signature (in this case I<b0>). If it does not exist,
2834 a new I<Bucket List> is created for our key (and the next 15 future keys that
2835 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2836 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2837 this point, unless we are replacing an existing I<Bucket>), where the actual
2838 data will be stored.
2842 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2843 (or index number), then walking along the indexes. If there are enough
2844 keys/elements in this DB level, there might be nested indexes, each linked to
2845 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2846 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2847 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2848 plain key are stored.
2850 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2851 methods. In this process the indexes are walked systematically, and each key
2852 fetched in increasing MD5 order (which is why it appears random). Once the
2853 I<Bucket> is found, the value is skipped and the plain key returned instead.
2854 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2855 alphabetically sorted. This only happens on an index-level -- as soon as the
2856 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2857 so it's pretty much undefined how the keys will come out -- just like Perl's
2860 =head1 CODE COVERAGE
2862 We use B<Devel::Cover> to test the code coverage of our tests, below is the
2863 B<Devel::Cover> report on this module's test suite.
2865 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2866 File stmt bran cond sub pod time total
2867 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2868 blib/lib/DBM/Deep.pm 95.2 83.8 70.0 98.2 100.0 58.0 91.0
2869 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 n/a 26.7 98.0
2870 blib/lib/DBM/Deep/Hash.pm 95.3 80.0 100.0 100.0 n/a 15.3 92.4
2871 Total 96.2 84.8 74.4 98.8 100.0 100.0 92.4
2872 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2874 =head1 MORE INFORMATION
2876 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
2877 or send email to L<DBM-Deep@googlegroups.com>.
2881 Joseph Huckaby, L<jhuckaby@cpan.org>
2883 Rob Kinyon, L<rkinyon@cpan.org>
2885 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2889 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2890 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2894 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
2895 This is free software, you may use it and distribute it under the
2896 same terms as Perl itself.