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_NULL () { 'N' }
94 sub SIG_DATA () { 'D' }
95 sub SIG_INDEX () { 'I' }
96 sub SIG_BLIST () { 'B' }
100 # Setup constants for users to pass to new()
102 sub TYPE_HASH () { return SIG_HASH; }
103 sub TYPE_ARRAY () { return SIG_ARRAY; }
107 # Class constructor method for Perl OO interface.
108 # Calls tie() and returns blessed reference to tied hash or array,
109 # providing a hybrid OO/tie interface.
113 if (scalar(@_) > 1) { $args = {@_}; }
114 else { $args = { file => shift }; }
117 # Check if we want a tied hash or array.
120 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
121 $class = 'DBM::Deep::Array';
122 require DBM::Deep::Array;
123 tie @$self, $class, %$args;
126 $class = 'DBM::Deep::Hash';
127 require DBM::Deep::Hash;
128 tie %$self, $class, %$args;
131 return bless $self, $class;
136 # Setup $self and bless into this class.
141 # These are the defaults to be optionally overridden below
144 base_offset => length(SIG_FILE),
147 foreach my $param ( keys %$self ) {
148 next unless exists $args->{$param};
149 $self->{$param} = delete $args->{$param}
152 $self->{root} = exists $args->{root}
154 : DBM::Deep::_::Root->new( $args );
156 if (!defined($self->fh)) { $self->_open(); }
163 require DBM::Deep::Hash;
164 return DBM::Deep::Hash->TIEHASH( @_ );
169 require DBM::Deep::Array;
170 return DBM::Deep::Array->TIEARRAY( @_ );
173 #XXX Unneeded now ...
179 # Open a FileHandle to the database, create if nonexistent.
180 # Make sure file signature matches DeepDB spec.
182 my $self = $_[0]->_get_self;
184 if (defined($self->fh)) { $self->_close(); }
187 # Theoretically, adding O_BINARY should remove the need for the binmode
188 # Of course, testing it is going to be ... interesting.
189 my $flags = O_RDWR | O_CREAT | O_BINARY;
191 #XXX Can the mode be anything but r+, w+, or a+??
192 #XXX ie, it has to be in read-write mode
193 #XXX So, should we verify that the mode is legitimate?
195 #XXX Maybe the mode thingy should just go away. There's no good
196 #XXX reason for it ...
197 if ( $self->root->{mode} eq 'w+' ) {
202 sysopen( $fh, $self->root->{file}, $flags )
204 $self->root->{fh} = $fh;
205 }; if ($@ ) { $self->_throw_error( "Received error: $@\n" ); }
206 if (! defined($self->fh)) {
207 return $self->_throw_error("Cannot sysopen file: " . $self->root->{file} . ": $!");
212 #XXX Can we remove this by using the right sysopen() flags?
213 # Maybe ... q.v. above
214 binmode $fh; # for win32
216 if ($self->root->{autoflush}) {
217 my $old = select $fh;
223 seek($fh, 0, SEEK_SET);
226 my $bytes_read = read( $fh, $signature, length(SIG_FILE));
229 # File is empty -- write signature and master index
232 seek($fh, 0, SEEK_SET);
234 $self->root->{end} = length(SIG_FILE);
235 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
237 my $plain_key = "[base]";
238 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
239 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
241 # Flush the filehandle
242 my $old_fh = select $fh;
252 # Check signature was valid
254 unless ($signature eq SIG_FILE) {
256 return $self->_throw_error("Signature not found -- file is not a Deep DB");
259 $self->root->{end} = (stat($fh))[7];
262 # Get our type from master index signature
264 my $tag = $self->_load_tag($self->base_offset);
266 #XXX We probably also want to store the hash algorithm name and not assume anything
267 #XXX The cool thing would be to allow a different hashing algorithm at every level
270 return $self->_throw_error("Corrupted file, no master index record");
272 if ($self->{type} ne $tag->{signature}) {
273 return $self->_throw_error("File type mismatch");
281 # Close database FileHandle
283 my $self = $_[0]->_get_self;
284 close $self->root->{fh};
289 # Given offset, signature and content, create tag and write to disk
291 my ($self, $offset, $sig, $content) = @_;
292 my $size = length($content);
296 seek($fh, $offset, SEEK_SET);
297 print($fh $sig . pack($DATA_LENGTH_PACK, $size) . $content );
299 if ($offset == $self->root->{end}) {
300 $self->root->{end} += SIG_SIZE + $DATA_LENGTH_SIZE + $size;
306 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
313 # Given offset, load single tag and return signature, size and data
320 seek($fh, $offset, SEEK_SET);
321 if (eof $fh) { return undef; }
324 read( $fh, $sig, SIG_SIZE);
327 read( $fh, $size, $DATA_LENGTH_SIZE);
328 $size = unpack($DATA_LENGTH_PACK, $size);
331 read( $fh, $buffer, $size);
336 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
343 # Given index tag, lookup single entry in index and return .
346 my ($tag, $index) = @_;
348 my $location = unpack($LONG_PACK, substr($tag->{content}, $index * $LONG_SIZE, $LONG_SIZE) );
349 if (!$location) { return; }
351 return $self->_load_tag( $location );
356 # Adds one key/value pair to bucket list, given offset, MD5 digest of key,
357 # plain (undigested) key and value.
360 my ($tag, $md5, $plain_key, $value) = @_;
361 my $keys = $tag->{content};
365 my $is_dbm_deep = eval { $value->isa( 'DBM::Deep' ) };
366 my $internal_ref = $is_dbm_deep && ($value->root eq $self->root);
371 # Iterate through buckets, seeing if this is a new entry or a replace.
373 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
374 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
375 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
378 # Found empty bucket (end of list). Populate and exit loop.
382 $location = $internal_ref
383 ? $value->base_offset
384 : $self->root->{end};
386 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
387 print($fh $md5 . pack($LONG_PACK, $location) );
390 elsif ($md5 eq $key) {
392 # Found existing bucket with same key. Replace with new value.
397 $location = $value->base_offset;
398 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
399 print($fh $md5 . pack($LONG_PACK, $location) );
402 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
404 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
407 # If value is a hash, array, or raw value with equal or less size, we can
408 # reuse the same content area of the database. Otherwise, we have to create
409 # a new content area at the EOF.
412 my $r = Scalar::Util::reftype( $value ) || '';
413 if ( $r eq 'HASH' || $r eq 'ARRAY' ) { $actual_length = $INDEX_SIZE; }
414 else { $actual_length = length($value); }
416 if ($actual_length <= $size) {
420 $location = $self->root->{end};
421 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE, SEEK_SET);
422 print($fh pack($LONG_PACK, $location) );
430 # If this is an internal reference, return now.
431 # No need to write value or plain key
438 # If bucket didn't fit into list, split into a new index level
441 seek($fh, $tag->{ref_loc}, SEEK_SET);
442 print($fh pack($LONG_PACK, $self->root->{end}) );
444 my $index_tag = $self->_create_tag($self->root->{end}, SIG_INDEX, chr(0) x $INDEX_SIZE);
447 $keys .= $md5 . pack($LONG_PACK, 0);
449 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
450 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
452 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
453 my $num = ord(substr($key, $tag->{ch} + 1, 1));
455 if ($offsets[$num]) {
456 my $offset = $offsets[$num] + SIG_SIZE + $DATA_LENGTH_SIZE;
457 seek($fh, $offset, SEEK_SET);
459 read( $fh, $subkeys, $BUCKET_LIST_SIZE);
461 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
462 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
464 seek($fh, $offset + ($k * $BUCKET_SIZE), SEEK_SET);
465 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
471 $offsets[$num] = $self->root->{end};
472 seek($fh, $index_tag->{offset} + ($num * $LONG_SIZE), SEEK_SET);
473 print($fh pack($LONG_PACK, $self->root->{end}) );
475 my $blist_tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
477 seek($fh, $blist_tag->{offset}, SEEK_SET);
478 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
483 $location ||= $self->root->{end};
484 } # re-index bucket list
487 # Seek to content area and store signature, value and plaintext key
491 seek($fh, $location, SEEK_SET);
494 # Write signature based on content type, set content length and write actual value.
496 my $r = Scalar::Util::reftype($value) || '';
498 print($fh TYPE_HASH );
499 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
500 $content_length = $INDEX_SIZE;
502 elsif ($r eq 'ARRAY') {
503 print($fh TYPE_ARRAY );
504 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
505 $content_length = $INDEX_SIZE;
507 elsif (!defined($value)) {
508 print($fh SIG_NULL );
509 print($fh pack($DATA_LENGTH_PACK, 0) );
513 print($fh SIG_DATA );
514 print($fh pack($DATA_LENGTH_PACK, length($value)) . $value );
515 $content_length = length($value);
519 # Plain key is stored AFTER value, as keys are typically fetched less often.
521 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
524 # If value is blessed, preserve class name
526 if ( $self->root->{autobless} ) {
527 my $value_class = Scalar::Util::blessed($value);
528 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
530 # Blessed ref -- will restore later
533 print($fh pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
534 $content_length += 1;
535 $content_length += $DATA_LENGTH_SIZE + length($value_class);
539 $content_length += 1;
544 # If this is a new content area, advance EOF counter
546 if ($location == $self->root->{end}) {
547 $self->root->{end} += SIG_SIZE;
548 $self->root->{end} += $DATA_LENGTH_SIZE + $content_length;
549 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
553 # If content is a hash or array, create new child DeepDB object and
554 # pass each key or element to it.
557 my $branch = DBM::Deep->new(
559 base_offset => $location,
562 foreach my $key (keys %{$value}) {
563 #$branch->{$key} = $value->{$key};
564 $branch->STORE( $key, $value->{$key} );
567 elsif ($r eq 'ARRAY') {
568 my $branch = DBM::Deep->new(
570 base_offset => $location,
574 foreach my $element (@{$value}) {
575 #$branch->[$index] = $element;
576 $branch->STORE( $index, $element );
584 return $self->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
587 sub _get_bucket_value {
589 # Fetch single value given tag and MD5 digested key.
592 my ($tag, $md5) = @_;
593 my $keys = $tag->{content};
598 # Iterate through buckets, looking for a key match
601 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
602 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
603 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
607 # Hit end of list, no match
612 if ( $md5 ne $key ) {
617 # Found match -- seek to offset and read signature
620 seek($fh, $subloc, SEEK_SET);
621 read( $fh, $signature, SIG_SIZE);
624 # If value is a hash or array, return new DeepDB object with correct offset
626 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
627 my $obj = DBM::Deep->new(
629 base_offset => $subloc,
633 if ($self->root->{autobless}) {
635 # Skip over value and plain key to see if object needs
638 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
641 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
642 if ($size) { seek($fh, $size, SEEK_CUR); }
645 read( $fh, $bless_bit, 1);
646 if (ord($bless_bit)) {
648 # Yes, object needs to be re-blessed
651 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
652 if ($size) { read( $fh, $class_name, $size); }
653 if ($class_name) { $obj = bless( $obj, $class_name ); }
661 # Otherwise return actual value
663 elsif ($signature eq SIG_DATA) {
666 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
667 if ($size) { read( $fh, $value, $size); }
672 # Key exists, but content is null
682 # Delete single key/value pair given tag and MD5 digested key.
685 my ($tag, $md5) = @_;
686 my $keys = $tag->{content};
691 # Iterate through buckets, looking for a key match
694 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
695 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
696 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
700 # Hit end of list, no match
705 if ( $md5 ne $key ) {
710 # Matched key -- delete bucket and return
712 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
713 print($fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
714 print($fh chr(0) x $BUCKET_SIZE );
724 # Check existence of single key given tag and MD5 digested key.
727 my ($tag, $md5) = @_;
728 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 -- return true
758 sub _find_bucket_list {
760 # Locate offset for bucket list, given digested key
766 # Locate offset for bucket list using digest index system
769 my $tag = $self->_load_tag($self->base_offset);
770 if (!$tag) { return; }
772 while ($tag->{signature} ne SIG_BLIST) {
773 $tag = $self->_index_lookup($tag, ord(substr($md5, $ch, 1)));
774 if (!$tag) { return; }
781 sub _traverse_index {
783 # Scan index and recursively step into deeper levels, looking for next key.
785 my ($self, $offset, $ch, $force_return_next) = @_;
786 $force_return_next = undef unless $force_return_next;
788 my $tag = $self->_load_tag( $offset );
792 if ($tag->{signature} ne SIG_BLIST) {
793 my $content = $tag->{content};
795 if ($self->{return_next}) { $start = 0; }
796 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
798 for (my $index = $start; $index < 256; $index++) {
799 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
801 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
802 if (defined($result)) { return $result; }
806 $self->{return_next} = 1;
809 elsif ($tag->{signature} eq SIG_BLIST) {
810 my $keys = $tag->{content};
811 if ($force_return_next) { $self->{return_next} = 1; }
814 # Iterate through buckets, looking for a key match
816 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
817 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
818 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
822 # End of bucket list -- return to outer loop
824 $self->{return_next} = 1;
827 elsif ($key eq $self->{prev_md5}) {
829 # Located previous key -- return next one found
831 $self->{return_next} = 1;
834 elsif ($self->{return_next}) {
836 # Seek to bucket location and skip over signature
838 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
841 # Skip over value to get to plain key
844 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
845 if ($size) { seek($fh, $size, SEEK_CUR); }
848 # Read in plain key and return as scalar
851 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
852 if ($size) { read( $fh, $plain_key, $size); }
858 $self->{return_next} = 1;
859 } # tag is a bucket list
866 # Locate next key, given digested previous one
868 my $self = $_[0]->_get_self;
870 $self->{prev_md5} = $_[1] ? $_[1] : undef;
871 $self->{return_next} = 0;
874 # If the previous key was not specifed, start at the top and
875 # return the first one found.
877 if (!$self->{prev_md5}) {
878 $self->{prev_md5} = chr(0) x $HASH_SIZE;
879 $self->{return_next} = 1;
882 return $self->_traverse_index( $self->base_offset, 0 );
887 # If db locking is set, flock() the db file. If called multiple
888 # times before unlock(), then the same number of unlocks() must
889 # be called before the lock is released.
891 my $self = $_[0]->_get_self;
893 $type = LOCK_EX unless defined $type;
895 if ($self->root->{locking}) {
896 if (!$self->root->{locked}) { flock($self->fh, $type); }
897 $self->root->{locked}++;
907 # If db locking is set, unlock the db file. See note in lock()
908 # regarding calling lock() multiple times.
910 my $self = $_[0]->_get_self;
912 if ($self->root->{locking} && $self->root->{locked} > 0) {
913 $self->root->{locked}--;
914 if (!$self->root->{locked}) { flock($self->fh, LOCK_UN); }
922 #XXX These uses of ref() need verified
925 # Copy single level of keys or elements to new DB handle.
926 # Recurse for nested structures
928 my $self = $_[0]->_get_self;
931 if ($self->type eq TYPE_HASH) {
932 my $key = $self->first_key();
934 my $value = $self->get($key);
935 #XXX This doesn't work with autobless
936 if (!ref($value)) { $db_temp->{$key} = $value; }
938 my $type = $value->type;
939 if ($type eq TYPE_HASH) { $db_temp->{$key} = {}; }
940 else { $db_temp->{$key} = []; }
941 $value->_copy_node( $db_temp->{$key} );
943 $key = $self->next_key($key);
947 my $length = $self->length();
948 for (my $index = 0; $index < $length; $index++) {
949 my $value = $self->get($index);
950 if (!ref($value)) { $db_temp->[$index] = $value; }
951 #XXX NO tests for this code
953 my $type = $value->type;
954 if ($type eq TYPE_HASH) { $db_temp->[$index] = {}; }
955 else { $db_temp->[$index] = []; }
956 $value->_copy_node( $db_temp->[$index] );
964 # Recursively export into standard Perl hashes and arrays.
966 my $self = $_[0]->_get_self;
969 if ($self->type eq TYPE_HASH) { $temp = {}; }
970 elsif ($self->type eq TYPE_ARRAY) { $temp = []; }
973 $self->_copy_node( $temp );
981 # Recursively import Perl hash/array structure
983 #XXX This use of ref() seems to be ok
984 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
986 my $self = $_[0]->_get_self;
989 #XXX This use of ref() seems to be ok
992 # struct is not a reference, so just import based on our type
996 if ($self->type eq TYPE_HASH) { $struct = {@_}; }
997 elsif ($self->type eq TYPE_ARRAY) { $struct = [@_]; }
1000 my $r = Scalar::Util::reftype($struct) || '';
1001 if ($r eq "HASH" && $self->type eq TYPE_HASH) {
1002 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
1004 elsif ($r eq "ARRAY" && $self->type eq TYPE_ARRAY) {
1005 $self->push( @$struct );
1008 return $self->_throw_error("Cannot import: type mismatch");
1016 # Rebuild entire database into new file, then move
1017 # it back on top of original.
1019 my $self = $_[0]->_get_self;
1021 #XXX Need to create a new test for this
1022 # if ($self->root->{links} > 1) {
1023 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
1026 my $db_temp = DBM::Deep->new(
1027 file => $self->root->{file} . '.tmp',
1031 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
1035 $self->_copy_node( $db_temp );
1039 # Attempt to copy user, group and permissions over to new file
1041 my @stats = stat($self->fh);
1042 my $perms = $stats[2] & 07777;
1043 my $uid = $stats[4];
1044 my $gid = $stats[5];
1045 chown( $uid, $gid, $self->root->{file} . '.tmp' );
1046 chmod( $perms, $self->root->{file} . '.tmp' );
1048 # q.v. perlport for more information on this variable
1049 if ( $^O eq 'MSWin32' ) {
1051 # Potential race condition when optmizing on Win32 with locking.
1052 # The Windows filesystem requires that the filehandle be closed
1053 # before it is overwritten with rename(). This could be redone
1060 if (!rename $self->root->{file} . '.tmp', $self->root->{file}) {
1061 unlink $self->root->{file} . '.tmp';
1063 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
1075 # Make copy of object and return
1077 my $self = $_[0]->_get_self;
1079 return DBM::Deep->new(
1080 type => $self->type,
1081 base_offset => $self->base_offset,
1087 my %is_legal_filter = map {
1090 store_key store_value
1091 fetch_key fetch_value
1096 # Setup filter function for storing or fetching the key or value
1098 my $self = $_[0]->_get_self;
1099 my $type = lc $_[1];
1100 my $func = $_[2] ? $_[2] : undef;
1102 if ( $is_legal_filter{$type} ) {
1103 $self->root->{"filter_$type"} = $func;
1117 # Get access to the root structure
1119 my $self = $_[0]->_get_self;
1120 return $self->{root};
1125 # Get access to the raw FileHandle
1127 #XXX It will be useful, though, when we split out HASH and ARRAY
1128 my $self = $_[0]->_get_self;
1129 return $self->root->{fh};
1134 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
1136 my $self = $_[0]->_get_self;
1137 return $self->{type};
1142 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
1144 my $self = $_[0]->_get_self;
1145 return $self->{base_offset};
1150 # Get last error string, or undef if no error
1153 #? ( _get_self($_[0])->{root}->{error} or undef )
1154 ? ( $_[0]->_get_self->{root}->{error} or undef )
1164 # Store error string in self
1166 my $self = $_[0]->_get_self;
1167 my $error_text = $_[1];
1169 $self->root->{error} = $error_text;
1171 unless ($self->root->{debug}) {
1172 die "DBM::Deep: $error_text\n";
1175 warn "DBM::Deep: $error_text\n";
1183 my $self = $_[0]->_get_self;
1185 undef $self->root->{error};
1190 # Precalculate index, bucket and bucket list sizes
1193 #XXX I don't like this ...
1194 set_pack() unless defined $LONG_SIZE;
1196 $INDEX_SIZE = 256 * $LONG_SIZE;
1197 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
1198 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
1203 # Set pack/unpack modes (see file header for more)
1205 my ($long_s, $long_p, $data_s, $data_p) = @_;
1207 $LONG_SIZE = $long_s ? $long_s : 4;
1208 $LONG_PACK = $long_p ? $long_p : 'N';
1210 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
1211 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
1218 # Set key digest function (default is MD5)
1220 my ($digest_func, $hash_size) = @_;
1222 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
1223 $HASH_SIZE = $hash_size ? $hash_size : 16;
1229 # tie() methods (hashes and arrays)
1234 # Store single hash key/value or array element in database.
1236 my $self = $_[0]->_get_self;
1237 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1238 #XXX What is ref() checking here?
1239 #YYY User may be storing a hash, in which case we do not want it run
1240 #YYY through the filtering system
1241 my $value = ($self->root->{filter_store_value} && !ref($_[2])) ? $self->root->{filter_store_value}->($_[2]) : $_[2];
1243 my $unpacked_key = $key;
1244 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1245 my $md5 = $DIGEST_FUNC->($key);
1248 # Make sure file is open
1250 if (!defined($self->fh) && !$self->_open()) {
1258 # Request exclusive lock for writing
1260 $self->lock( LOCK_EX );
1263 # If locking is enabled, set 'end' parameter again, in case another
1264 # DB instance appended to our file while we were unlocked.
1266 if ($self->root->{locking} || $self->root->{volatile}) {
1267 $self->root->{end} = (stat($fh))[7];
1271 # Locate offset for bucket list using digest index system
1273 my $tag = $self->_load_tag($self->base_offset);
1275 $tag = $self->_create_tag($self->base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1279 while ($tag->{signature} ne SIG_BLIST) {
1280 my $num = ord(substr($md5, $ch, 1));
1281 my $new_tag = $self->_index_lookup($tag, $num);
1283 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1284 seek($fh, $ref_loc, SEEK_SET);
1285 print($fh pack($LONG_PACK, $self->root->{end}) );
1287 $tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1288 $tag->{ref_loc} = $ref_loc;
1293 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1295 $tag->{ref_loc} = $ref_loc;
1302 # Add key/value to bucket list
1304 my $result = $self->_add_bucket( $tag, $md5, $key, $value );
1307 # If this object is an array, and bucket was not a replace, and key is numerical,
1308 # and index is equal or greater than current length, advance length variable.
1310 if (($result == 2) && ($self->type eq TYPE_ARRAY) && ($unpacked_key =~ /^\d+$/) && ($unpacked_key >= $self->FETCHSIZE())) {
1311 $self->STORESIZE( $unpacked_key + 1 );
1321 # Fetch single value or element given plain key or array index
1323 my $self = $_[0]->_get_self;
1326 if ( $self->type eq TYPE_HASH ) {
1327 if ( my $filter = $self->root->{filter_store_key} ) {
1328 $key = $filter->( $key );
1331 elsif ( $self->type eq TYPE_ARRAY ) {
1332 if ( $key =~ /^\d+$/ ) {
1333 $key = pack($LONG_PACK, $key);
1337 my $md5 = $DIGEST_FUNC->($key);
1340 # Make sure file is open
1342 if (!defined($self->fh)) { $self->_open(); }
1345 # Request shared lock for reading
1347 $self->lock( LOCK_SH );
1349 my $tag = $self->_find_bucket_list( $md5 );
1356 # Get value from bucket list
1358 my $result = $self->_get_bucket_value( $tag, $md5 );
1362 #XXX What is ref() checking here?
1363 return ($result && !ref($result) && $self->root->{filter_fetch_value}) ? $self->root->{filter_fetch_value}->($result) : $result;
1368 # Delete single key/value pair or element given plain key or array index
1370 my $self = $_[0]->_get_self;
1371 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1373 my $unpacked_key = $key;
1374 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1375 my $md5 = $DIGEST_FUNC->($key);
1378 # Make sure file is open
1380 if (!defined($self->fh)) { $self->_open(); }
1383 # Request exclusive lock for writing
1385 $self->lock( LOCK_EX );
1387 my $tag = $self->_find_bucket_list( $md5 );
1396 my $result = $self->_delete_bucket( $tag, $md5 );
1399 # If this object is an array and the key deleted was on the end of the stack,
1400 # decrement the length variable.
1402 if ($result && ($self->type eq TYPE_ARRAY) && ($unpacked_key == $self->FETCHSIZE() - 1)) {
1403 $self->STORESIZE( $unpacked_key );
1413 # Check if a single key or element exists given plain key or array index
1415 my $self = $_[0]->_get_self;
1416 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1418 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1419 my $md5 = $DIGEST_FUNC->($key);
1422 # Make sure file is open
1424 if (!defined($self->fh)) { $self->_open(); }
1427 # Request shared lock for reading
1429 $self->lock( LOCK_SH );
1431 my $tag = $self->_find_bucket_list( $md5 );
1434 # For some reason, the built-in exists() function returns '' for false
1442 # Check if bucket exists and return 1 or ''
1444 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1453 # Clear all keys from hash, or all elements from array.
1455 my $self = $_[0]->_get_self;
1458 # Make sure file is open
1460 if (!defined($self->fh)) { $self->_open(); }
1463 # Request exclusive lock for writing
1465 $self->lock( LOCK_EX );
1469 seek($fh, $self->base_offset, SEEK_SET);
1475 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
1483 # Public method aliases
1485 *put = *store = *STORE;
1486 *get = *fetch = *FETCH;
1491 package DBM::Deep::_::Root;
1506 filter_store_key => undef,
1507 filter_store_value => undef,
1508 filter_fetch_key => undef,
1509 filter_fetch_value => undef,
1520 return unless $self;
1522 close $self->{fh} if $self->{fh};
1533 DBM::Deep - A pure perl multi-level hash/array DBM
1538 my $db = DBM::Deep->new( "foo.db" );
1540 $db->{key} = 'value'; # tie() style
1543 $db->put('key', 'value'); # OO style
1544 print $db->get('key');
1546 # true multi-level support
1547 $db->{my_complex} = [
1548 'hello', { perl => 'rules' },
1553 A unique flat-file database module, written in pure perl. True
1554 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1555 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1556 handle millions of keys and unlimited hash levels without significant
1557 slow-down. Written from the ground-up in pure perl -- this is NOT a
1558 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1559 Mac OS X and Windows.
1563 Hopefully you are using CPAN's excellent Perl module, which will download
1564 and install the module for you. If not, get the tarball, and run these
1576 Construction can be done OO-style (which is the recommended way), or using
1577 Perl's tie() function. Both are examined here.
1579 =head2 OO CONSTRUCTION
1581 The recommended way to construct a DBM::Deep object is to use the new()
1582 method, which gets you a blessed, tied hash or array reference.
1584 my $db = DBM::Deep->new( "foo.db" );
1586 This opens a new database handle, mapped to the file "foo.db". If this
1587 file does not exist, it will automatically be created. DB files are
1588 opened in "r+" (read/write) mode, and the type of object returned is a
1589 hash, unless otherwise specified (see L<OPTIONS> below).
1593 You can pass a number of options to the constructor to specify things like
1594 locking, autoflush, etc. This is done by passing an inline hash:
1596 my $db = DBM::Deep->new(
1602 Notice that the filename is now specified I<inside> the hash with
1603 the "file" parameter, as opposed to being the sole argument to the
1604 constructor. This is required if any options are specified.
1605 See L<OPTIONS> below for the complete list.
1609 You can also start with an array instead of a hash. For this, you must
1610 specify the C<type> parameter:
1612 my $db = DBM::Deep->new(
1614 type => DBM::Deep->TYPE_ARRAY
1617 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1618 a new DB file. If you create a DBM::Deep object with an existing file, the
1619 C<type> will be loaded from the file header, and ignored if it is passed
1622 =head2 TIE CONSTRUCTION
1624 Alternatively, you can create a DBM::Deep handle by using Perl's built-in
1625 tie() function. This is not ideal, because you get only a basic, tied hash
1626 (or array) which is not blessed, so you can't call any functions on it.
1629 tie %hash, "DBM::Deep", "foo.db";
1632 tie @array, "DBM::Deep", "bar.db";
1634 As with the OO constructor, you can replace the DB filename parameter with
1635 a hash containing one or more options (see L<OPTIONS> just below for the
1638 tie %hash, "DBM::Deep", {
1646 There are a number of options that can be passed in when constructing your
1647 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1653 Filename of the DB file to link the handle to. You can pass a full absolute
1654 filesystem path, partial path, or a plain filename if the file is in the
1655 current working directory. This is a required parameter.
1659 File open mode (read-only, read-write, etc.) string passed to Perl's FileHandle
1660 module. This is an optional parameter, and defaults to "r+" (read/write).
1661 B<Note:> If the default (r+) mode is selected, the file will also be auto-
1662 created if it doesn't exist.
1666 This parameter specifies what type of object to create, a hash or array. Use
1667 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1668 This only takes effect when beginning a new file. This is an optional
1669 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1673 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1674 function to lock the database in exclusive mode for writes, and shared mode for
1675 reads. Pass any true value to enable. This affects the base DB handle I<and
1676 any child hashes or arrays> that use the same DB file. This is an optional
1677 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1681 Specifies whether autoflush is to be enabled on the underlying FileHandle.
1682 This obviously slows down write operations, but is required if you may have
1683 multiple processes accessing the same DB file (also consider enable I<locking>
1684 or at least I<volatile>). Pass any true value to enable. This is an optional
1685 parameter, and defaults to 0 (disabled).
1689 If I<volatile> mode is enabled, DBM::Deep will stat() the DB file before each
1690 STORE() operation. This is required if an outside force may change the size of
1691 the file between transactions. Locking also implicitly enables volatile. This
1692 is useful if you want to use a different locking system or write your own. Pass
1693 any true value to enable. This is an optional parameter, and defaults to 0
1698 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1699 restore them when fetched. This is an B<experimental> feature, and does have
1700 side-effects. Basically, when hashes are re-blessed into their original
1701 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1702 able to call any DBM::Deep methods on them. You have been warned.
1703 This is an optional parameter, and defaults to 0 (disabled).
1707 See L<FILTERS> below.
1711 Setting I<debug> mode will make all errors non-fatal, dump them out to
1712 STDERR, and continue on. This is for debugging purposes only, and probably
1713 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1717 Instead of passing a file path, you can instead pass a handle to an pre-opened
1718 filehandle. Note: Beware of using the magick *DATA handle, as this actually
1719 contains your entire Perl script, as well as the data following the __DATA__
1720 marker. This will not work, because DBM::Deep uses absolute seek()s into the
1721 file. Instead, consider reading *DATA into an IO::Scalar handle, then passing
1726 =head1 TIE INTERFACE
1728 With DBM::Deep you can access your databases using Perl's standard hash/array
1729 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can treat
1730 them as such. DBM::Deep will intercept all reads/writes and direct them to the right
1731 place -- the DB file. This has nothing to do with the L<TIE CONSTRUCTION>
1732 section above. This simply tells you how to use DBM::Deep using regular hashes
1733 and arrays, rather than calling functions like C<get()> and C<put()> (although those
1734 work too). It is entirely up to you how to want to access your databases.
1738 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1739 or even nested hashes (or arrays) using standard Perl syntax:
1741 my $db = DBM::Deep->new( "foo.db" );
1743 $db->{mykey} = "myvalue";
1745 $db->{myhash}->{subkey} = "subvalue";
1747 print $db->{myhash}->{subkey} . "\n";
1749 You can even step through hash keys using the normal Perl C<keys()> function:
1751 foreach my $key (keys %$db) {
1752 print "$key: " . $db->{$key} . "\n";
1755 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1756 pushes them onto an array, all before the loop even begins. If you have an
1757 extra large hash, this may exhaust Perl's memory. Instead, consider using
1758 Perl's C<each()> function, which pulls keys/values one at a time, using very
1761 while (my ($key, $value) = each %$db) {
1762 print "$key: $value\n";
1765 Please note that when using C<each()>, you should always pass a direct
1766 hash reference, not a lookup. Meaning, you should B<never> do this:
1769 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1771 This causes an infinite loop, because for each iteration, Perl is calling
1772 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1773 it effectively keeps returning the first key over and over again. Instead,
1774 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1778 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1779 reference. This includes inserting, removing and manipulating elements,
1780 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1781 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1782 or simply be a nested array reference inside a hash. Example:
1784 my $db = DBM::Deep->new(
1785 file => "foo-array.db",
1786 type => DBM::Deep->TYPE_ARRAY
1790 push @$db, "bar", "baz";
1791 unshift @$db, "bah";
1793 my $last_elem = pop @$db; # baz
1794 my $first_elem = shift @$db; # bah
1795 my $second_elem = $db->[1]; # bar
1797 my $num_elements = scalar @$db;
1801 In addition to the I<tie()> interface, you can also use a standard OO interface
1802 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1803 array) has its own methods, but both types share the following common methods:
1804 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1810 Stores a new hash key/value pair, or sets an array element value. Takes two
1811 arguments, the hash key or array index, and the new value. The value can be
1812 a scalar, hash ref or array ref. Returns true on success, false on failure.
1814 $db->put("foo", "bar"); # for hashes
1815 $db->put(1, "bar"); # for arrays
1819 Fetches the value of a hash key or array element. Takes one argument: the hash
1820 key or array index. Returns a scalar, hash ref or array ref, depending on the
1823 my $value = $db->get("foo"); # for hashes
1824 my $value = $db->get(1); # for arrays
1828 Checks if a hash key or array index exists. Takes one argument: the hash key
1829 or array index. Returns true if it exists, false if not.
1831 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1832 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1836 Deletes one hash key/value pair or array element. Takes one argument: the hash
1837 key or array index. Returns true on success, false if not found. For arrays,
1838 the remaining elements located after the deleted element are NOT moved over.
1839 The deleted element is essentially just undefined, which is exactly how Perl's
1840 internal arrays work. Please note that the space occupied by the deleted
1841 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1842 below for details and workarounds.
1844 $db->delete("foo"); # for hashes
1845 $db->delete(1); # for arrays
1849 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1850 value. Please note that the space occupied by the deleted keys/values or
1851 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1852 details and workarounds.
1854 $db->clear(); # hashes or arrays
1860 For hashes, DBM::Deep supports all the common methods described above, and the
1861 following additional methods: C<first_key()> and C<next_key()>.
1867 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1868 fetched in an undefined order (which appears random). Takes no arguments,
1869 returns the key as a scalar value.
1871 my $key = $db->first_key();
1875 Returns the "next" key in the hash, given the previous one as the sole argument.
1876 Returns undef if there are no more keys to be fetched.
1878 $key = $db->next_key($key);
1882 Here are some examples of using hashes:
1884 my $db = DBM::Deep->new( "foo.db" );
1886 $db->put("foo", "bar");
1887 print "foo: " . $db->get("foo") . "\n";
1889 $db->put("baz", {}); # new child hash ref
1890 $db->get("baz")->put("buz", "biz");
1891 print "buz: " . $db->get("baz")->get("buz") . "\n";
1893 my $key = $db->first_key();
1895 print "$key: " . $db->get($key) . "\n";
1896 $key = $db->next_key($key);
1899 if ($db->exists("foo")) { $db->delete("foo"); }
1903 For arrays, DBM::Deep supports all the common methods described above, and the
1904 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1905 C<unshift()> and C<splice()>.
1911 Returns the number of elements in the array. Takes no arguments.
1913 my $len = $db->length();
1917 Adds one or more elements onto the end of the array. Accepts scalars, hash
1918 refs or array refs. No return value.
1920 $db->push("foo", "bar", {});
1924 Fetches the last element in the array, and deletes it. Takes no arguments.
1925 Returns undef if array is empty. Returns the element value.
1927 my $elem = $db->pop();
1931 Fetches the first element in the array, deletes it, then shifts all the
1932 remaining elements over to take up the space. Returns the element value. This
1933 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1936 my $elem = $db->shift();
1940 Inserts one or more elements onto the beginning of the array, shifting all
1941 existing elements over to make room. Accepts scalars, hash refs or array refs.
1942 No return value. This method is not recommended with large arrays -- see
1943 <LARGE ARRAYS> below for details.
1945 $db->unshift("foo", "bar", {});
1949 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1950 -f splice> for usage -- it is too complicated to document here. This method is
1951 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1955 Here are some examples of using arrays:
1957 my $db = DBM::Deep->new(
1959 type => DBM::Deep->TYPE_ARRAY
1962 $db->push("bar", "baz");
1963 $db->unshift("foo");
1966 my $len = $db->length();
1967 print "length: $len\n"; # 4
1969 for (my $k=0; $k<$len; $k++) {
1970 print "$k: " . $db->get($k) . "\n";
1973 $db->splice(1, 2, "biz", "baf");
1975 while (my $elem = shift @$db) {
1976 print "shifted: $elem\n";
1981 Enable automatic file locking by passing a true value to the C<locking>
1982 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1984 my $db = DBM::Deep->new(
1989 This causes DBM::Deep to C<flock()> the underlying FileHandle object with exclusive
1990 mode for writes, and shared mode for reads. This is required if you have
1991 multiple processes accessing the same database file, to avoid file corruption.
1992 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1993 NFS> below for more.
1995 =head2 EXPLICIT LOCKING
1997 You can explicitly lock a database, so it remains locked for multiple
1998 transactions. This is done by calling the C<lock()> method, and passing an
1999 optional lock mode argument (defaults to exclusive mode). This is particularly
2000 useful for things like counters, where the current value needs to be fetched,
2001 then incremented, then stored again.
2004 my $counter = $db->get("counter");
2006 $db->put("counter", $counter);
2015 You can pass C<lock()> an optional argument, which specifies which mode to use
2016 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2017 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2018 same as the constants defined in Perl's C<Fcntl> module.
2020 $db->lock( DBM::Deep->LOCK_SH );
2024 If you want to implement your own file locking scheme, be sure to create your
2025 DBM::Deep objects setting the C<volatile> option to true. This hints to DBM::Deep
2026 that the DB file may change between transactions. See L<LOW-LEVEL ACCESS>
2029 =head1 IMPORTING/EXPORTING
2031 You can import existing complex structures by calling the C<import()> method,
2032 and export an entire database into an in-memory structure using the C<export()>
2033 method. Both are examined here.
2037 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2038 walking the structure and adding keys/elements to the database as you go,
2039 simply pass a reference to the C<import()> method. This recursively adds
2040 everything to an existing DBM::Deep object for you. Here is an example:
2045 array1 => [ "elem0", "elem1", "elem2" ],
2047 subkey1 => "subvalue1",
2048 subkey2 => "subvalue2"
2052 my $db = DBM::Deep->new( "foo.db" );
2053 $db->import( $struct );
2055 print $db->{key1} . "\n"; # prints "value1"
2057 This recursively imports the entire C<$struct> object into C<$db>, including
2058 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2059 keys are merged with the existing ones, replacing if they already exist.
2060 The C<import()> method can be called on any database level (not just the base
2061 level), and works with both hash and array DB types.
2065 B<Note:> Make sure your existing structure has no circular references in it.
2066 These will cause an infinite loop when importing.
2070 Calling the C<export()> method on an existing DBM::Deep object will return
2071 a reference to a new in-memory copy of the database. The export is done
2072 recursively, so all nested hashes/arrays are all exported to standard Perl
2073 objects. Here is an example:
2075 my $db = DBM::Deep->new( "foo.db" );
2077 $db->{key1} = "value1";
2078 $db->{key2} = "value2";
2080 $db->{hash1}->{subkey1} = "subvalue1";
2081 $db->{hash1}->{subkey2} = "subvalue2";
2083 my $struct = $db->export();
2085 print $struct->{key1} . "\n"; # prints "value1"
2087 This makes a complete copy of the database in memory, and returns a reference
2088 to it. The C<export()> method can be called on any database level (not just
2089 the base level), and works with both hash and array DB types. Be careful of
2090 large databases -- you can store a lot more data in a DBM::Deep object than an
2091 in-memory Perl structure.
2095 B<Note:> Make sure your database has no circular references in it.
2096 These will cause an infinite loop when exporting.
2100 DBM::Deep has a number of hooks where you can specify your own Perl function
2101 to perform filtering on incoming or outgoing data. This is a perfect
2102 way to extend the engine, and implement things like real-time compression or
2103 encryption. Filtering applies to the base DB level, and all child hashes /
2104 arrays. Filter hooks can be specified when your DBM::Deep object is first
2105 constructed, or by calling the C<set_filter()> method at any time. There are
2106 four available filter hooks, described below:
2110 =item * filter_store_key
2112 This filter is called whenever a hash key is stored. It
2113 is passed the incoming key, and expected to return a transformed key.
2115 =item * filter_store_value
2117 This filter is called whenever a hash key or array element is stored. It
2118 is passed the incoming value, and expected to return a transformed value.
2120 =item * filter_fetch_key
2122 This filter is called whenever a hash key is fetched (i.e. via
2123 C<first_key()> or C<next_key()>). It is passed the transformed key,
2124 and expected to return the plain key.
2126 =item * filter_fetch_value
2128 This filter is called whenever a hash key or array element is fetched.
2129 It is passed the transformed value, and expected to return the plain value.
2133 Here are the two ways to setup a filter hook:
2135 my $db = DBM::Deep->new(
2137 filter_store_value => \&my_filter_store,
2138 filter_fetch_value => \&my_filter_fetch
2143 $db->set_filter( "filter_store_value", \&my_filter_store );
2144 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2146 Your filter function will be called only when dealing with SCALAR keys or
2147 values. When nested hashes and arrays are being stored/fetched, filtering
2148 is bypassed. Filters are called as static functions, passed a single SCALAR
2149 argument, and expected to return a single SCALAR value. If you want to
2150 remove a filter, set the function reference to C<undef>:
2152 $db->set_filter( "filter_store_value", undef );
2154 =head2 REAL-TIME ENCRYPTION EXAMPLE
2156 Here is a working example that uses the I<Crypt::Blowfish> module to
2157 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2158 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2159 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2162 use Crypt::Blowfish;
2165 my $cipher = Crypt::CBC->new({
2166 'key' => 'my secret key',
2167 'cipher' => 'Blowfish',
2169 'regenerate_key' => 0,
2170 'padding' => 'space',
2174 my $db = DBM::Deep->new(
2175 file => "foo-encrypt.db",
2176 filter_store_key => \&my_encrypt,
2177 filter_store_value => \&my_encrypt,
2178 filter_fetch_key => \&my_decrypt,
2179 filter_fetch_value => \&my_decrypt,
2182 $db->{key1} = "value1";
2183 $db->{key2} = "value2";
2184 print "key1: " . $db->{key1} . "\n";
2185 print "key2: " . $db->{key2} . "\n";
2191 return $cipher->encrypt( $_[0] );
2194 return $cipher->decrypt( $_[0] );
2197 =head2 REAL-TIME COMPRESSION EXAMPLE
2199 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2200 compression / decompression of keys & values with DBM::Deep Filters.
2201 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2202 more on I<Compress::Zlib>.
2207 my $db = DBM::Deep->new(
2208 file => "foo-compress.db",
2209 filter_store_key => \&my_compress,
2210 filter_store_value => \&my_compress,
2211 filter_fetch_key => \&my_decompress,
2212 filter_fetch_value => \&my_decompress,
2215 $db->{key1} = "value1";
2216 $db->{key2} = "value2";
2217 print "key1: " . $db->{key1} . "\n";
2218 print "key2: " . $db->{key2} . "\n";
2224 return Compress::Zlib::memGzip( $_[0] ) ;
2227 return Compress::Zlib::memGunzip( $_[0] ) ;
2230 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2231 actually numerical index numbers, and are not filtered.
2233 =head1 ERROR HANDLING
2235 Most DBM::Deep methods return a true value for success, and call die() on
2236 failure. You can wrap calls in an eval block to catch the die. Also, the
2237 actual error message is stored in an internal scalar, which can be fetched by
2238 calling the C<error()> method.
2240 my $db = DBM::Deep->new( "foo.db" ); # create hash
2241 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2243 print $db->error(); # prints error message
2245 You can then call C<clear_error()> to clear the current error state.
2249 If you set the C<debug> option to true when creating your DBM::Deep object,
2250 all errors are considered NON-FATAL, and dumped to STDERR. This is only
2251 for debugging purposes.
2253 =head1 LARGEFILE SUPPORT
2255 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2256 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2257 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2258 by calling the static C<set_pack()> method before you do anything else.
2260 DBM::Deep::set_pack(8, 'Q');
2262 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2263 instead of 32-bit longs. After setting these values your DB files have a
2264 theoretical maximum size of 16 XB (exabytes).
2268 B<Note:> Changing these values will B<NOT> work for existing database files.
2269 Only change this for new files, and make sure it stays set consistently
2270 throughout the file's life. If you do set these values, you can no longer
2271 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2272 back to 32-bit mode.
2276 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2277 only a 32-bit Perl. However, I have received user reports that this does
2280 =head1 LOW-LEVEL ACCESS
2282 If you require low-level access to the underlying FileHandle that DBM::Deep uses,
2283 you can call the C<fh()> method, which returns the handle:
2287 This method can be called on the root level of the datbase, or any child
2288 hashes or arrays. All levels share a I<root> structure, which contains things
2289 like the FileHandle, a reference counter, and all your options you specified
2290 when you created the object. You can get access to this root structure by
2291 calling the C<root()> method.
2293 my $root = $db->root();
2295 This is useful for changing options after the object has already been created,
2296 such as enabling/disabling locking, volatile or debug modes. You can also
2297 store your own temporary user data in this structure (be wary of name
2298 collision), which is then accessible from any child hash or array.
2300 =head1 CUSTOM DIGEST ALGORITHM
2302 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2303 keys. However you can override this, and use another algorithm (such as SHA-256)
2304 or even write your own. But please note that DBM::Deep currently expects zero
2305 collisions, so your algorithm has to be I<perfect>, so to speak.
2306 Collision detection may be introduced in a later version.
2310 You can specify a custom digest algorithm by calling the static C<set_digest()>
2311 function, passing a reference to a subroutine, and the length of the algorithm's
2312 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2313 objects. Here is a working example that uses a 256-bit hash from the
2314 I<Digest::SHA256> module. Please see
2315 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2320 my $context = Digest::SHA256::new(256);
2322 DBM::Deep::set_digest( \&my_digest, 32 );
2324 my $db = DBM::Deep->new( "foo-sha.db" );
2326 $db->{key1} = "value1";
2327 $db->{key2} = "value2";
2328 print "key1: " . $db->{key1} . "\n";
2329 print "key2: " . $db->{key2} . "\n";
2335 return substr( $context->hash($_[0]), 0, 32 );
2338 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2339 of bytes you specify in the C<set_digest()> function (in this case 32).
2341 =head1 CIRCULAR REFERENCES
2343 DBM::Deep has B<experimental> support for circular references. Meaning you
2344 can have a nested hash key or array element that points to a parent object.
2345 This relationship is stored in the DB file, and is preserved between sessions.
2348 my $db = DBM::Deep->new( "foo.db" );
2351 $db->{circle} = $db; # ref to self
2353 print $db->{foo} . "\n"; # prints "foo"
2354 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2356 One catch is, passing the object to a function that recursively walks the
2357 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2358 C<export()> methods) will result in an infinite loop. The other catch is,
2359 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2360 or C<next_key()> methods), you will get the I<target object's key>, not the
2361 ref's key. This gets even more interesting with the above example, where
2362 the I<circle> key points to the base DB object, which technically doesn't
2363 have a key. So I made DBM::Deep return "[base]" as the key name in that
2366 =head1 CAVEATS / ISSUES / BUGS
2368 This section describes all the known issues with DBM::Deep. It you have found
2369 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2371 =head2 UNUSED SPACE RECOVERY
2373 One major caveat with DBM::Deep is that space occupied by existing keys and
2374 values is not recovered when they are deleted. Meaning if you keep deleting
2375 and adding new keys, your file will continuously grow. I am working on this,
2376 but in the meantime you can call the built-in C<optimize()> method from time to
2377 time (perhaps in a crontab or something) to recover all your unused space.
2379 $db->optimize(); # returns true on success
2381 This rebuilds the ENTIRE database into a new file, then moves it on top of
2382 the original. The new file will have no unused space, thus it will take up as
2383 little disk space as possible. Please note that this operation can take
2384 a long time for large files, and you need enough disk space to temporarily hold
2385 2 copies of your DB file. The temporary file is created in the same directory
2386 as the original, named with a ".tmp" extension, and is deleted when the
2387 operation completes. Oh, and if locking is enabled, the DB is automatically
2388 locked for the entire duration of the copy.
2392 B<WARNING:> Only call optimize() on the top-level node of the database, and
2393 make sure there are no child references lying around. DBM::Deep keeps a reference
2394 counter, and if it is greater than 1, optimize() will abort and return undef.
2396 =head2 AUTOVIVIFICATION
2398 Unfortunately, autovivification doesn't work with tied hashes. This appears to
2399 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
2400 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
2401 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
2402 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
2405 $db->{foo}->{bar} = "hello";
2407 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
2408 being an empty hash. Try this instead, which works fine:
2410 $db->{foo} = { bar => "hello" };
2412 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
2413 the execution path, and Perl indeed passes an empty hash to the STORE() method.
2414 Probably a bug in Perl.
2416 =head2 FILE CORRUPTION
2418 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2419 for a 32-bit signature when opened, but other corruption in files can cause
2420 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2421 stuck in an infinite loop depending on the level of corruption. File write
2422 operations are not checked for failure (for speed), so if you happen to run
2423 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2424 be addressed in a later version of DBM::Deep.
2428 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2429 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2430 about setting up your NFS server with a locking daemon, then using lockf() to
2431 lock your files, but your milage may vary there as well. From what I
2432 understand, there is no real way to do it. However, if you need access to the
2433 underlying FileHandle in DBM::Deep for using some other kind of locking scheme like
2434 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2436 =head2 COPYING OBJECTS
2438 Beware of copying tied objects in Perl. Very strange things can happen.
2439 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2440 returns a new, blessed, tied hash or array to the same level in the DB.
2442 my $copy = $db->clone();
2446 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2447 These functions cause every element in the array to move, which can be murder
2448 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2449 a different location. This may be addressed in a later version.
2453 This section discusses DBM::Deep's speed and memory usage.
2457 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2458 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2459 multi-level hash/array support, and cross-platform FTPable files. Even so,
2460 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2461 with huge databases. Here is some test data:
2463 Adding 1,000,000 keys to new DB file...
2465 At 100 keys, avg. speed is 2,703 keys/sec
2466 At 200 keys, avg. speed is 2,642 keys/sec
2467 At 300 keys, avg. speed is 2,598 keys/sec
2468 At 400 keys, avg. speed is 2,578 keys/sec
2469 At 500 keys, avg. speed is 2,722 keys/sec
2470 At 600 keys, avg. speed is 2,628 keys/sec
2471 At 700 keys, avg. speed is 2,700 keys/sec
2472 At 800 keys, avg. speed is 2,607 keys/sec
2473 At 900 keys, avg. speed is 2,190 keys/sec
2474 At 1,000 keys, avg. speed is 2,570 keys/sec
2475 At 2,000 keys, avg. speed is 2,417 keys/sec
2476 At 3,000 keys, avg. speed is 1,982 keys/sec
2477 At 4,000 keys, avg. speed is 1,568 keys/sec
2478 At 5,000 keys, avg. speed is 1,533 keys/sec
2479 At 6,000 keys, avg. speed is 1,787 keys/sec
2480 At 7,000 keys, avg. speed is 1,977 keys/sec
2481 At 8,000 keys, avg. speed is 2,028 keys/sec
2482 At 9,000 keys, avg. speed is 2,077 keys/sec
2483 At 10,000 keys, avg. speed is 2,031 keys/sec
2484 At 20,000 keys, avg. speed is 1,970 keys/sec
2485 At 30,000 keys, avg. speed is 2,050 keys/sec
2486 At 40,000 keys, avg. speed is 2,073 keys/sec
2487 At 50,000 keys, avg. speed is 1,973 keys/sec
2488 At 60,000 keys, avg. speed is 1,914 keys/sec
2489 At 70,000 keys, avg. speed is 2,091 keys/sec
2490 At 80,000 keys, avg. speed is 2,103 keys/sec
2491 At 90,000 keys, avg. speed is 1,886 keys/sec
2492 At 100,000 keys, avg. speed is 1,970 keys/sec
2493 At 200,000 keys, avg. speed is 2,053 keys/sec
2494 At 300,000 keys, avg. speed is 1,697 keys/sec
2495 At 400,000 keys, avg. speed is 1,838 keys/sec
2496 At 500,000 keys, avg. speed is 1,941 keys/sec
2497 At 600,000 keys, avg. speed is 1,930 keys/sec
2498 At 700,000 keys, avg. speed is 1,735 keys/sec
2499 At 800,000 keys, avg. speed is 1,795 keys/sec
2500 At 900,000 keys, avg. speed is 1,221 keys/sec
2501 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2503 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2504 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2505 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2506 Run time was 12 min 3 sec.
2510 One of the great things about DBM::Deep is that it uses very little memory.
2511 Even with huge databases (1,000,000+ keys) you will not see much increased
2512 memory on your process. DBM::Deep relies solely on the filesystem for storing
2513 and fetching data. Here is output from I</usr/bin/top> before even opening a
2516 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2517 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2519 Basically the process is taking 2,716K of memory. And here is the same
2520 process after storing and fetching 1,000,000 keys:
2522 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2523 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2525 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2526 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2528 =head1 DB FILE FORMAT
2530 In case you were interested in the underlying DB file format, it is documented
2531 here in this section. You don't need to know this to use the module, it's just
2532 included for reference.
2536 DBM::Deep files always start with a 32-bit signature to identify the file type.
2537 This is at offset 0. The signature is "DPDB" in network byte order. This is
2538 checked when the file is opened.
2542 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2543 has a standard header containing the type of data, the length of data, and then
2544 the data itself. The type is a single character (1 byte), the length is a
2545 32-bit unsigned long in network byte order, and the data is, well, the data.
2546 Here is how it unfolds:
2550 Immediately after the 32-bit file signature is the I<Master Index> record.
2551 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2552 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2553 depending on how the DBM::Deep object was constructed.
2557 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2558 number). The first 8-bit char of the MD5 signature is the offset into the
2559 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2560 index element is a file offset of the next tag for the key/element in question,
2561 which is usually a I<Bucket List> tag (see below).
2565 The next tag I<could> be another index, depending on how many keys/elements
2566 exist. See L<RE-INDEXING> below for details.
2570 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2571 file offsets to where the actual data is stored. It starts with a standard
2572 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2573 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2574 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2575 When the list fills up, a I<Re-Index> operation is performed (See
2576 L<RE-INDEXING> below).
2580 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2581 index/value pair (in array mode). It starts with a standard tag header with
2582 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2583 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2584 header. The size reported in the tag header is only for the value, but then,
2585 just after the value is another size (32-bit unsigned long) and then the plain
2586 key itself. Since the value is likely to be fetched more often than the plain
2587 key, I figured it would be I<slightly> faster to store the value first.
2591 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2592 record for the nested structure, where the process begins all over again.
2596 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2597 exhausted. Then, when another key/element comes in, the list is converted to a
2598 new index record. However, this index will look at the next char in the MD5
2599 hash, and arrange new Bucket List pointers accordingly. This process is called
2600 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2601 17 (16 + new one) keys/elements are removed from the old Bucket List and
2602 inserted into the new index. Several new Bucket Lists are created in the
2603 process, as a new MD5 char from the key is being examined (it is unlikely that
2604 the keys will all share the same next char of their MD5s).
2608 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2609 when the Bucket Lists will turn into indexes, but the first round tends to
2610 happen right around 4,000 keys. You will see a I<slight> decrease in
2611 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2612 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2613 right around 900,000 keys. This process can continue nearly indefinitely --
2614 right up until the point the I<MD5> signatures start colliding with each other,
2615 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2616 getting struck by lightning while you are walking to cash in your tickets.
2617 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2618 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2619 this is 340 unodecillion, but don't quote me).
2623 When a new key/element is stored, the key (or index number) is first ran through
2624 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2625 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2626 for the first char of the signature (in this case I<b>). If it does not exist,
2627 a new I<Bucket List> is created for our key (and the next 15 future keys that
2628 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2629 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2630 this point, unless we are replacing an existing I<Bucket>), where the actual
2631 data will be stored.
2635 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2636 (or index number), then walking along the indexes. If there are enough
2637 keys/elements in this DB level, there might be nested indexes, each linked to
2638 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2639 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2640 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2641 plain key are stored.
2645 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2646 methods. In this process the indexes are walked systematically, and each key
2647 fetched in increasing MD5 order (which is why it appears random). Once the
2648 I<Bucket> is found, the value is skipped the plain key returned instead.
2649 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2650 alphabetically sorted. This only happens on an index-level -- as soon as the
2651 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2652 so it's pretty much undefined how the keys will come out -- just like Perl's
2655 =head1 CODE COVERAGE
2657 I use B<Devel::Cover> to test the code coverage of my tests, below is the B<Devel::Cover> report on this
2658 module's test suite.
2660 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2661 File stmt bran cond sub pod time total
2662 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2663 blib/lib/DBM/Deep.pm 93.9 82.4 74.7 97.9 10.5 85.7 88.0
2664 blib/lib/DBM/Deep/Array.pm 97.8 84.6 50.0 100.0 n/a 9.0 94.6
2665 blib/lib/DBM/Deep/Hash.pm 93.9 87.5 100.0 100.0 n/a 5.3 93.4
2666 Total 94.4 82.9 75.8 98.5 10.5 100.0 89.0
2667 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2671 Joseph Huckaby, L<jhuckaby@cpan.org>
2673 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2677 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2678 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2682 Copyright (c) 2002-2005 Joseph Huckaby. All Rights Reserved.
2683 This is free software, you may use it and distribute it under the
2684 same terms as Perl itself.