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 " . $db->{my_complex}->[1]->{perl} . "!\n";
27 # (c) 2002-2005 Joseph Huckaby. All Rights Reserved.
28 # This program is free software; you can redistribute it and/or
29 # modify it under the same terms as Perl itself.
34 use Fcntl qw(:DEFAULT :flock :seek);
37 use vars qw/$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;
135 my @outer_params = qw( type base_offset );
138 # Setup $self and bless into this class.
145 base_offset => length(SIG_FILE),
150 foreach my $outer_parm ( @outer_params ) {
151 next unless exists $args->{$outer_parm};
152 $self->{$outer_parm} = delete $args->{$outer_parm}
155 $self->{root} = exists $args->{root}
157 : DBM::Deep::_::Root->new( $args );
159 if (!defined($self->fh)) { $self->_open(); }
166 tied( %{$_[0]} ) || $_[0]
171 require DBM::Deep::Hash;
172 return DBM::Deep::Hash->TIEHASH( @_ );
177 require DBM::Deep::Array;
178 return DBM::Deep::Array->TIEARRAY( @_ );
181 #XXX Unneeded now ...
185 my %translate_mode = (
195 # Open a FileHandle to the database, create if nonexistent.
196 # Make sure file signature matches DeepDB spec.
198 my $self = _get_self($_[0]);
200 if (defined($self->fh)) { $self->_close(); }
203 my $filename = $self->root->{file};
204 #XXX Can the mode be anything but r+, w+, or a+??
205 #XXX ie, it has to be in read-write mode
206 my $mode = $translate_mode{ $self->root->{mode} };
208 if (!(-e $filename) && $mode eq '+<') {
209 sysopen( FH, $filename, O_CREAT | O_WRONLY, 0666 );
214 sysopen( $fh, $filename, O_RDWR )
216 $self->root->{fh} = $fh;
217 }; if ($@ ) { $self->_throw_error( "Received error: $@\n" ); }
218 if (! defined($self->fh)) {
219 return $self->_throw_error("Cannot sysopen file: " . $self->root->{file} . ": $!");
224 #XXX Can we remove this by using the right sysopen() flags?
225 #XXX I don't think so - there's an item in fopen(3) about rb+, but I'm not sure
227 binmode $fh; # for win32
229 if ($self->root->{autoflush}) {
230 my $old = select $fh;
236 seek($fh, 0, SEEK_SET);
237 my $bytes_read = read( $fh, $signature, length(SIG_FILE));
240 # File is empty -- write signature and master index
243 seek($fh, 0, SEEK_SET);
245 $self->root->{end} = length(SIG_FILE);
246 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
248 my $plain_key = "[base]";
249 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
250 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
252 # Flush the filehandle
253 my $old_fh = select $fh;
263 # Check signature was valid
265 unless ($signature eq SIG_FILE) {
267 return $self->_throw_error("Signature not found -- file is not a Deep DB");
270 $self->root->{end} = (stat($fh))[7];
273 # Get our type from master index signature
275 my $tag = $self->_load_tag($self->base_offset);
277 #XXX We probably also want to store the hash algorithm name and not assume anything
280 return $self->_throw_error("Corrupted file, no master index record");
282 if ($self->{type} ne $tag->{signature}) {
283 return $self->_throw_error("File type mismatch");
291 # Close database FileHandle
293 my $self = _get_self($_[0]);
294 close $self->root->{fh};
299 # Given offset, signature and content, create tag and write to disk
301 my ($self, $offset, $sig, $content) = @_;
302 my $size = length($content);
306 seek($fh, $offset, SEEK_SET);
307 print($fh $sig . pack($DATA_LENGTH_PACK, $size) . $content );
309 if ($offset == $self->root->{end}) {
310 $self->root->{end} += SIG_SIZE + $DATA_LENGTH_SIZE + $size;
316 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
323 # Given offset, load single tag and return signature, size and data
330 seek($fh, $offset, SEEK_SET);
331 if (eof $fh) { return undef; }
334 read( $fh, $sig, SIG_SIZE);
337 read( $fh, $size, $DATA_LENGTH_SIZE);
338 $size = unpack($DATA_LENGTH_PACK, $size);
341 read( $fh, $buffer, $size);
346 offset => $offset + SIG_SIZE + $DATA_LENGTH_SIZE,
353 # Given index tag, lookup single entry in index and return .
356 my ($tag, $index) = @_;
358 my $location = unpack($LONG_PACK, substr($tag->{content}, $index * $LONG_SIZE, $LONG_SIZE) );
359 if (!$location) { return; }
361 return $self->_load_tag( $location );
366 # Adds one key/value pair to bucket list, given offset, MD5 digest of key,
367 # plain (undigested) key and value.
370 my ($tag, $md5, $plain_key, $value) = @_;
371 my $keys = $tag->{content};
375 my $is_dbm_deep = eval { $value->isa( 'DBM::Deep' ) };
376 my $internal_ref = $is_dbm_deep && ($value->root eq $self->root);
381 # Iterate through buckets, seeing if this is a new entry or a replace.
383 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
384 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
385 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
388 # Found empty bucket (end of list). Populate and exit loop.
392 $location = $internal_ref
393 ? $value->base_offset
394 : $self->root->{end};
396 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
397 print($fh $md5 . pack($LONG_PACK, $location) );
400 elsif ($md5 eq $key) {
402 # Found existing bucket with same key. Replace with new value.
407 $location = $value->base_offset;
408 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
409 print($fh $md5 . pack($LONG_PACK, $location) );
412 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
414 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
417 # If value is a hash, array, or raw value with equal or less size, we can
418 # reuse the same content area of the database. Otherwise, we have to create
419 # a new content area at the EOF.
422 my $r = Scalar::Util::reftype( $value ) || '';
423 if ( $r eq 'HASH' || $r eq 'ARRAY' ) { $actual_length = $INDEX_SIZE; }
424 else { $actual_length = length($value); }
426 if ($actual_length <= $size) {
430 $location = $self->root->{end};
431 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE, SEEK_SET);
432 print($fh pack($LONG_PACK, $location) );
440 # If this is an internal reference, return now.
441 # No need to write value or plain key
448 # If bucket didn't fit into list, split into a new index level
451 seek($fh, $tag->{ref_loc}, SEEK_SET);
452 print($fh pack($LONG_PACK, $self->root->{end}) );
454 my $index_tag = $self->_create_tag($self->root->{end}, SIG_INDEX, chr(0) x $INDEX_SIZE);
457 $keys .= $md5 . pack($LONG_PACK, 0);
459 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
460 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
462 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
463 my $num = ord(substr($key, $tag->{ch} + 1, 1));
465 if ($offsets[$num]) {
466 my $offset = $offsets[$num] + SIG_SIZE + $DATA_LENGTH_SIZE;
467 seek($fh, $offset, SEEK_SET);
469 read( $fh, $subkeys, $BUCKET_LIST_SIZE);
471 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
472 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
474 seek($fh, $offset + ($k * $BUCKET_SIZE), SEEK_SET);
475 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
481 $offsets[$num] = $self->root->{end};
482 seek($fh, $index_tag->{offset} + ($num * $LONG_SIZE), SEEK_SET);
483 print($fh pack($LONG_PACK, $self->root->{end}) );
485 my $blist_tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
487 seek($fh, $blist_tag->{offset}, SEEK_SET);
488 print($fh $key . pack($LONG_PACK, $old_subloc || $self->root->{end}) );
493 $location ||= $self->root->{end};
494 } # re-index bucket list
497 # Seek to content area and store signature, value and plaintext key
501 seek($fh, $location, SEEK_SET);
504 # Write signature based on content type, set content length and write actual value.
506 my $r = Scalar::Util::reftype($value) || '';
508 print($fh TYPE_HASH );
509 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
510 $content_length = $INDEX_SIZE;
512 elsif ($r eq 'ARRAY') {
513 print($fh TYPE_ARRAY );
514 print($fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
515 $content_length = $INDEX_SIZE;
517 elsif (!defined($value)) {
518 print($fh SIG_NULL );
519 print($fh pack($DATA_LENGTH_PACK, 0) );
523 print($fh SIG_DATA );
524 print($fh pack($DATA_LENGTH_PACK, length($value)) . $value );
525 $content_length = length($value);
529 # Plain key is stored AFTER value, as keys are typically fetched less often.
531 print($fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
534 # If value is blessed, preserve class name
536 if ( $self->root->{autobless} ) {
537 my $value_class = Scalar::Util::blessed($value);
538 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
540 # Blessed ref -- will restore later
543 print($fh pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
544 $content_length += 1;
545 $content_length += $DATA_LENGTH_SIZE + length($value_class);
549 $content_length += 1;
554 # If this is a new content area, advance EOF counter
556 if ($location == $self->root->{end}) {
557 $self->root->{end} += SIG_SIZE;
558 $self->root->{end} += $DATA_LENGTH_SIZE + $content_length;
559 $self->root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
563 # If content is a hash or array, create new child DeepDB object and
564 # pass each key or element to it.
567 my $branch = DBM::Deep->new(
569 base_offset => $location,
572 foreach my $key (keys %{$value}) {
573 #$branch->{$key} = $value->{$key};
574 $branch->STORE( $key, $value->{$key} );
577 elsif ($r eq 'ARRAY') {
578 my $branch = DBM::Deep->new(
580 base_offset => $location,
584 foreach my $element (@{$value}) {
585 #$branch->[$index] = $element;
586 $branch->STORE( $index, $element );
594 return $self->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
597 sub _get_bucket_value {
599 # Fetch single value given tag and MD5 digested key.
602 my ($tag, $md5) = @_;
603 my $keys = $tag->{content};
608 # Iterate through buckets, looking for a key match
611 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
612 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
613 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
617 # Hit end of list, no match
622 if ( $md5 ne $key ) {
627 # Found match -- seek to offset and read signature
630 seek($fh, $subloc, SEEK_SET);
631 read( $fh, $signature, SIG_SIZE);
634 # If value is a hash or array, return new DeepDB object with correct offset
636 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
637 my $obj = DBM::Deep->new(
639 base_offset => $subloc,
643 if ($self->root->{autobless}) {
645 # Skip over value and plain key to see if object needs
648 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
651 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
652 if ($size) { seek($fh, $size, SEEK_CUR); }
655 read( $fh, $bless_bit, 1);
656 if (ord($bless_bit)) {
658 # Yes, object needs to be re-blessed
661 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
662 if ($size) { read( $fh, $class_name, $size); }
663 if ($class_name) { $obj = bless( $obj, $class_name ); }
671 # Otherwise return actual value
673 elsif ($signature eq SIG_DATA) {
676 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
677 if ($size) { read( $fh, $value, $size); }
682 # Key exists, but content is null
692 # Delete single key/value pair given tag and MD5 digested key.
695 my ($tag, $md5) = @_;
696 my $keys = $tag->{content};
701 # Iterate through buckets, looking for a key match
704 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
705 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
706 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
710 # Hit end of list, no match
715 if ( $md5 ne $key ) {
720 # Matched key -- delete bucket and return
722 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE), SEEK_SET);
723 print($fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
724 print($fh chr(0) x $BUCKET_SIZE );
734 # Check existence of single key given tag and MD5 digested key.
737 my ($tag, $md5) = @_;
738 my $keys = $tag->{content};
741 # Iterate through buckets, looking for a key match
744 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
745 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
746 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
750 # Hit end of list, no match
755 if ( $md5 ne $key ) {
760 # Matched key -- return true
768 sub _find_bucket_list {
770 # Locate offset for bucket list, given digested key
776 # Locate offset for bucket list using digest index system
779 my $tag = $self->_load_tag($self->base_offset);
780 if (!$tag) { return; }
782 while ($tag->{signature} ne SIG_BLIST) {
783 $tag = $self->_index_lookup($tag, ord(substr($md5, $ch, 1)));
784 if (!$tag) { return; }
791 sub _traverse_index {
793 # Scan index and recursively step into deeper levels, looking for next key.
795 my ($self, $offset, $ch, $force_return_next) = @_;
796 $force_return_next = undef unless $force_return_next;
798 my $tag = $self->_load_tag( $offset );
802 if ($tag->{signature} ne SIG_BLIST) {
803 my $content = $tag->{content};
805 if ($self->{return_next}) { $start = 0; }
806 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
808 for (my $index = $start; $index < 256; $index++) {
809 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
811 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
812 if (defined($result)) { return $result; }
816 $self->{return_next} = 1;
819 elsif ($tag->{signature} eq SIG_BLIST) {
820 my $keys = $tag->{content};
821 if ($force_return_next) { $self->{return_next} = 1; }
824 # Iterate through buckets, looking for a key match
826 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
827 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
828 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
832 # End of bucket list -- return to outer loop
834 $self->{return_next} = 1;
837 elsif ($key eq $self->{prev_md5}) {
839 # Located previous key -- return next one found
841 $self->{return_next} = 1;
844 elsif ($self->{return_next}) {
846 # Seek to bucket location and skip over signature
848 seek($fh, $subloc + SIG_SIZE, SEEK_SET);
851 # Skip over value to get to plain key
854 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
855 if ($size) { seek($fh, $size, SEEK_CUR); }
858 # Read in plain key and return as scalar
861 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
862 if ($size) { read( $fh, $plain_key, $size); }
868 $self->{return_next} = 1;
869 } # tag is a bucket list
876 # Locate next key, given digested previous one
878 my $self = _get_self($_[0]);
880 $self->{prev_md5} = $_[1] ? $_[1] : undef;
881 $self->{return_next} = 0;
884 # If the previous key was not specifed, start at the top and
885 # return the first one found.
887 if (!$self->{prev_md5}) {
888 $self->{prev_md5} = chr(0) x $HASH_SIZE;
889 $self->{return_next} = 1;
892 return $self->_traverse_index( $self->base_offset, 0 );
897 # If db locking is set, flock() the db file. If called multiple
898 # times before unlock(), then the same number of unlocks() must
899 # be called before the lock is released.
901 my $self = _get_self($_[0]);
903 $type = LOCK_EX unless defined $type;
905 if ($self->root->{locking}) {
906 if (!$self->root->{locked}) { flock($self->fh, $type); }
907 $self->root->{locked}++;
917 # If db locking is set, unlock the db file. See note in lock()
918 # regarding calling lock() multiple times.
920 my $self = _get_self($_[0]);
922 if ($self->root->{locking} && $self->root->{locked} > 0) {
923 $self->root->{locked}--;
924 if (!$self->root->{locked}) { flock($self->fh, LOCK_UN); }
932 #XXX These uses of ref() need verified
935 # Copy single level of keys or elements to new DB handle.
936 # Recurse for nested structures
938 my $self = _get_self($_[0]);
941 if ($self->type eq TYPE_HASH) {
942 my $key = $self->first_key();
944 my $value = $self->get($key);
945 #XXX This doesn't work with autobless
946 if (!ref($value)) { $db_temp->{$key} = $value; }
948 my $type = $value->type;
949 if ($type eq TYPE_HASH) { $db_temp->{$key} = {}; }
950 else { $db_temp->{$key} = []; }
951 $value->_copy_node( $db_temp->{$key} );
953 $key = $self->next_key($key);
957 my $length = $self->length();
958 for (my $index = 0; $index < $length; $index++) {
959 my $value = $self->get($index);
960 if (!ref($value)) { $db_temp->[$index] = $value; }
961 #XXX NO tests for this code
963 my $type = $value->type;
964 if ($type eq TYPE_HASH) { $db_temp->[$index] = {}; }
965 else { $db_temp->[$index] = []; }
966 $value->_copy_node( $db_temp->[$index] );
974 # Recursively export into standard Perl hashes and arrays.
976 my $self = _get_self($_[0]);
979 if ($self->type eq TYPE_HASH) { $temp = {}; }
980 elsif ($self->type eq TYPE_ARRAY) { $temp = []; }
983 $self->_copy_node( $temp );
991 # Recursively import Perl hash/array structure
993 #XXX This use of ref() seems to be ok
994 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
996 my $self = _get_self($_[0]);
999 #XXX This use of ref() seems to be ok
1000 if (!ref($struct)) {
1002 # struct is not a reference, so just import based on our type
1006 if ($self->type eq TYPE_HASH) { $struct = {@_}; }
1007 elsif ($self->type eq TYPE_ARRAY) { $struct = [@_]; }
1010 my $r = Scalar::Util::reftype($struct) || '';
1011 if ($r eq "HASH" && $self->type eq TYPE_HASH) {
1012 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
1014 elsif ($r eq "ARRAY" && $self->type eq TYPE_ARRAY) {
1015 $self->push( @$struct );
1018 return $self->_throw_error("Cannot import: type mismatch");
1026 # Rebuild entire database into new file, then move
1027 # it back on top of original.
1029 my $self = _get_self($_[0]);
1031 #XXX Need to create a new test for this
1032 # if ($self->root->{links} > 1) {
1033 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
1036 my $db_temp = DBM::Deep->new(
1037 file => $self->root->{file} . '.tmp',
1041 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
1045 $self->_copy_node( $db_temp );
1049 # Attempt to copy user, group and permissions over to new file
1051 my @stats = stat($self->fh);
1052 my $perms = $stats[2] & 07777;
1053 my $uid = $stats[4];
1054 my $gid = $stats[5];
1055 chown( $uid, $gid, $self->root->{file} . '.tmp' );
1056 chmod( $perms, $self->root->{file} . '.tmp' );
1058 # q.v. perlport for more information on this variable
1059 if ( $^O eq 'MSWin32' ) {
1061 # Potential race condition when optmizing on Win32 with locking.
1062 # The Windows filesystem requires that the filehandle be closed
1063 # before it is overwritten with rename(). This could be redone
1070 if (!rename $self->root->{file} . '.tmp', $self->root->{file}) {
1071 unlink $self->root->{file} . '.tmp';
1073 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
1085 # Make copy of object and return
1087 my $self = _get_self($_[0]);
1089 return DBM::Deep->new(
1090 type => $self->type,
1091 base_offset => $self->base_offset,
1097 my %is_legal_filter = map {
1100 store_key store_value
1101 fetch_key fetch_value
1106 # Setup filter function for storing or fetching the key or value
1108 my $self = _get_self($_[0]);
1109 my $type = lc $_[1];
1110 my $func = $_[2] ? $_[2] : undef;
1112 if ( $is_legal_filter{$type} ) {
1113 $self->root->{"filter_$type"} = $func;
1127 # Get access to the root structure
1129 my $self = _get_self($_[0]);
1130 return $self->{root};
1135 # Get access to the raw FileHandle
1137 #XXX It will be useful, though, when we split out HASH and ARRAY
1138 my $self = _get_self($_[0]);
1139 return $self->root->{fh};
1144 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
1146 my $self = _get_self($_[0]);
1147 return $self->{type};
1152 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
1154 my $self = _get_self($_[0]);
1155 return $self->{base_offset};
1160 # Get last error string, or undef if no error
1163 ? ( _get_self($_[0])->{root}->{error} or undef )
1173 # Store error string in self
1175 my $self = _get_self($_[0]);
1176 my $error_text = $_[1];
1178 $self->root->{error} = $error_text;
1180 unless ($self->root->{debug}) {
1181 die "DBM::Deep: $error_text\n";
1184 warn "DBM::Deep: $error_text\n";
1192 my $self = _get_self($_[0]);
1194 undef $self->root->{error};
1199 # Precalculate index, bucket and bucket list sizes
1202 #XXX I don't like this ...
1203 set_pack() unless defined $LONG_SIZE;
1205 $INDEX_SIZE = 256 * $LONG_SIZE;
1206 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
1207 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
1212 # Set pack/unpack modes (see file header for more)
1214 my ($long_s, $long_p, $data_s, $data_p) = @_;
1216 $LONG_SIZE = $long_s ? $long_s : 4;
1217 $LONG_PACK = $long_p ? $long_p : 'N';
1219 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
1220 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
1227 # Set key digest function (default is MD5)
1229 my ($digest_func, $hash_size) = @_;
1231 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
1232 $HASH_SIZE = $hash_size ? $hash_size : 16;
1238 # tie() methods (hashes and arrays)
1243 # Store single hash key/value or array element in database.
1245 my $self = _get_self($_[0]);
1246 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1247 #XXX What is ref() checking here?
1248 #YYY User may be storing a hash, in which case we do not want it run
1249 #YYY through the filtering system
1250 my $value = ($self->root->{filter_store_value} && !ref($_[2])) ? $self->root->{filter_store_value}->($_[2]) : $_[2];
1252 my $unpacked_key = $key;
1253 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1254 my $md5 = $DIGEST_FUNC->($key);
1257 # Make sure file is open
1259 if (!defined($self->fh) && !$self->_open()) {
1267 # Request exclusive lock for writing
1269 $self->lock( LOCK_EX );
1272 # If locking is enabled, set 'end' parameter again, in case another
1273 # DB instance appended to our file while we were unlocked.
1275 if ($self->root->{locking} || $self->root->{volatile}) {
1276 $self->root->{end} = (stat($fh))[7];
1280 # Locate offset for bucket list using digest index system
1282 my $tag = $self->_load_tag($self->base_offset);
1284 $tag = $self->_create_tag($self->base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
1288 while ($tag->{signature} ne SIG_BLIST) {
1289 my $num = ord(substr($md5, $ch, 1));
1290 my $new_tag = $self->_index_lookup($tag, $num);
1292 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1293 seek($fh, $ref_loc, SEEK_SET);
1294 print($fh pack($LONG_PACK, $self->root->{end}) );
1296 $tag = $self->_create_tag($self->root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
1297 $tag->{ref_loc} = $ref_loc;
1302 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
1304 $tag->{ref_loc} = $ref_loc;
1311 # Add key/value to bucket list
1313 my $result = $self->_add_bucket( $tag, $md5, $key, $value );
1316 # If this object is an array, and bucket was not a replace, and key is numerical,
1317 # and index is equal or greater than current length, advance length variable.
1319 if (($result == 2) && ($self->type eq TYPE_ARRAY) && ($unpacked_key =~ /^\d+$/) && ($unpacked_key >= $self->FETCHSIZE())) {
1320 $self->STORESIZE( $unpacked_key + 1 );
1330 # Fetch single value or element given plain key or array index
1332 my $self = _get_self($_[0]);
1335 if ( $self->type eq TYPE_HASH ) {
1336 if ( my $filter = $self->root->{filter_store_key} ) {
1337 $key = $filter->( $key );
1340 elsif ( $self->type eq TYPE_ARRAY ) {
1341 if ( $key =~ /^\d+$/ ) {
1342 $key = pack($LONG_PACK, $key);
1346 my $md5 = $DIGEST_FUNC->($key);
1349 # Make sure file is open
1351 if (!defined($self->fh)) { $self->_open(); }
1354 # Request shared lock for reading
1356 $self->lock( LOCK_SH );
1358 my $tag = $self->_find_bucket_list( $md5 );
1365 # Get value from bucket list
1367 my $result = $self->_get_bucket_value( $tag, $md5 );
1371 #XXX What is ref() checking here?
1372 return ($result && !ref($result) && $self->root->{filter_fetch_value}) ? $self->root->{filter_fetch_value}->($result) : $result;
1377 # Delete single key/value pair or element given plain key or array index
1379 my $self = _get_self($_[0]);
1380 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1382 my $unpacked_key = $key;
1383 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1384 my $md5 = $DIGEST_FUNC->($key);
1387 # Make sure file is open
1389 if (!defined($self->fh)) { $self->_open(); }
1392 # Request exclusive lock for writing
1394 $self->lock( LOCK_EX );
1396 my $tag = $self->_find_bucket_list( $md5 );
1405 my $result = $self->_delete_bucket( $tag, $md5 );
1408 # If this object is an array and the key deleted was on the end of the stack,
1409 # decrement the length variable.
1411 if ($result && ($self->type eq TYPE_ARRAY) && ($unpacked_key == $self->FETCHSIZE() - 1)) {
1412 $self->STORESIZE( $unpacked_key );
1422 # Check if a single key or element exists given plain key or array index
1424 my $self = _get_self($_[0]);
1425 my $key = ($self->root->{filter_store_key} && $self->type eq TYPE_HASH) ? $self->root->{filter_store_key}->($_[1]) : $_[1];
1427 if (($self->type eq TYPE_ARRAY) && ($key =~ /^\d+$/)) { $key = pack($LONG_PACK, $key); }
1428 my $md5 = $DIGEST_FUNC->($key);
1431 # Make sure file is open
1433 if (!defined($self->fh)) { $self->_open(); }
1436 # Request shared lock for reading
1438 $self->lock( LOCK_SH );
1440 my $tag = $self->_find_bucket_list( $md5 );
1443 # For some reason, the built-in exists() function returns '' for false
1451 # Check if bucket exists and return 1 or ''
1453 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1462 # Clear all keys from hash, or all elements from array.
1464 my $self = _get_self($_[0]);
1467 # Make sure file is open
1469 if (!defined($self->fh)) { $self->_open(); }
1472 # Request exclusive lock for writing
1474 $self->lock( LOCK_EX );
1478 seek($fh, $self->base_offset, SEEK_SET);
1484 $self->_create_tag($self->base_offset, $self->type, chr(0) x $INDEX_SIZE);
1492 # Public method aliases
1494 *put = *store = *STORE;
1495 *get = *fetch = *FETCH;
1500 package DBM::Deep::_::Root;
1515 filter_store_key => undef,
1516 filter_store_value => undef,
1517 filter_fetch_key => undef,
1518 filter_fetch_value => undef,
1529 return unless $self;
1531 close $self->{fh} if $self->{fh};
1542 DBM::Deep - A pure perl multi-level hash/array DBM
1547 my $db = DBM::Deep->new( "foo.db" );
1549 $db->{key} = 'value'; # tie() style
1552 $db->put('key', 'value'); # OO style
1553 print $db->get('key');
1555 # true multi-level support
1556 $db->{my_complex} = [
1557 'hello', { perl => 'rules' },
1562 A unique flat-file database module, written in pure perl. True
1563 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1564 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1565 handle millions of keys and unlimited hash levels without significant
1566 slow-down. Written from the ground-up in pure perl -- this is NOT a
1567 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1568 Mac OS X and Windows.
1572 Hopefully you are using CPAN's excellent Perl module, which will download
1573 and install the module for you. If not, get the tarball, and run these
1585 Construction can be done OO-style (which is the recommended way), or using
1586 Perl's tie() function. Both are examined here.
1588 =head2 OO CONSTRUCTION
1590 The recommended way to construct a DBM::Deep object is to use the new()
1591 method, which gets you a blessed, tied hash or array reference.
1593 my $db = DBM::Deep->new( "foo.db" );
1595 This opens a new database handle, mapped to the file "foo.db". If this
1596 file does not exist, it will automatically be created. DB files are
1597 opened in "r+" (read/write) mode, and the type of object returned is a
1598 hash, unless otherwise specified (see L<OPTIONS> below).
1602 You can pass a number of options to the constructor to specify things like
1603 locking, autoflush, etc. This is done by passing an inline hash:
1605 my $db = DBM::Deep->new(
1611 Notice that the filename is now specified I<inside> the hash with
1612 the "file" parameter, as opposed to being the sole argument to the
1613 constructor. This is required if any options are specified.
1614 See L<OPTIONS> below for the complete list.
1618 You can also start with an array instead of a hash. For this, you must
1619 specify the C<type> parameter:
1621 my $db = DBM::Deep->new(
1623 type => DBM::Deep->TYPE_ARRAY
1626 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1627 a new DB file. If you create a DBM::Deep object with an existing file, the
1628 C<type> will be loaded from the file header, and ignored if it is passed
1631 =head2 TIE CONSTRUCTION
1633 Alternatively, you can create a DBM::Deep handle by using Perl's built-in
1634 tie() function. This is not ideal, because you get only a basic, tied hash
1635 (or array) which is not blessed, so you can't call any functions on it.
1638 tie %hash, "DBM::Deep", "foo.db";
1641 tie @array, "DBM::Deep", "bar.db";
1643 As with the OO constructor, you can replace the DB filename parameter with
1644 a hash containing one or more options (see L<OPTIONS> just below for the
1647 tie %hash, "DBM::Deep", {
1655 There are a number of options that can be passed in when constructing your
1656 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1662 Filename of the DB file to link the handle to. You can pass a full absolute
1663 filesystem path, partial path, or a plain filename if the file is in the
1664 current working directory. This is a required parameter.
1668 File open mode (read-only, read-write, etc.) string passed to Perl's FileHandle
1669 module. This is an optional parameter, and defaults to "r+" (read/write).
1670 B<Note:> If the default (r+) mode is selected, the file will also be auto-
1671 created if it doesn't exist.
1675 This parameter specifies what type of object to create, a hash or array. Use
1676 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1677 This only takes effect when beginning a new file. This is an optional
1678 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1682 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1683 function to lock the database in exclusive mode for writes, and shared mode for
1684 reads. Pass any true value to enable. This affects the base DB handle I<and
1685 any child hashes or arrays> that use the same DB file. This is an optional
1686 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1690 Specifies whether autoflush is to be enabled on the underlying FileHandle.
1691 This obviously slows down write operations, but is required if you may have
1692 multiple processes accessing the same DB file (also consider enable I<locking>
1693 or at least I<volatile>). Pass any true value to enable. This is an optional
1694 parameter, and defaults to 0 (disabled).
1698 If I<volatile> mode is enabled, DBM::Deep will stat() the DB file before each
1699 STORE() operation. This is required if an outside force may change the size of
1700 the file between transactions. Locking also implicitly enables volatile. This
1701 is useful if you want to use a different locking system or write your own. Pass
1702 any true value to enable. This is an optional parameter, and defaults to 0
1707 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1708 restore them when fetched. This is an B<experimental> feature, and does have
1709 side-effects. Basically, when hashes are re-blessed into their original
1710 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1711 able to call any DBM::Deep methods on them. You have been warned.
1712 This is an optional parameter, and defaults to 0 (disabled).
1716 See L<FILTERS> below.
1720 Setting I<debug> mode will make all errors non-fatal, dump them out to
1721 STDERR, and continue on. This is for debugging purposes only, and probably
1722 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1726 Instead of passing a file path, you can instead pass a handle to an pre-opened
1727 filehandle. Note: Beware of using the magick *DATA handle, as this actually
1728 contains your entire Perl script, as well as the data following the __DATA__
1729 marker. This will not work, because DBM::Deep uses absolute seek()s into the
1730 file. Instead, consider reading *DATA into an IO::Scalar handle, then passing
1735 =head1 TIE INTERFACE
1737 With DBM::Deep you can access your databases using Perl's standard hash/array
1738 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can treat
1739 them as such. DBM::Deep will intercept all reads/writes and direct them to the right
1740 place -- the DB file. This has nothing to do with the L<TIE CONSTRUCTION>
1741 section above. This simply tells you how to use DBM::Deep using regular hashes
1742 and arrays, rather than calling functions like C<get()> and C<put()> (although those
1743 work too). It is entirely up to you how to want to access your databases.
1747 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1748 or even nested hashes (or arrays) using standard Perl syntax:
1750 my $db = DBM::Deep->new( "foo.db" );
1752 $db->{mykey} = "myvalue";
1754 $db->{myhash}->{subkey} = "subvalue";
1756 print $db->{myhash}->{subkey} . "\n";
1758 You can even step through hash keys using the normal Perl C<keys()> function:
1760 foreach my $key (keys %$db) {
1761 print "$key: " . $db->{$key} . "\n";
1764 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1765 pushes them onto an array, all before the loop even begins. If you have an
1766 extra large hash, this may exhaust Perl's memory. Instead, consider using
1767 Perl's C<each()> function, which pulls keys/values one at a time, using very
1770 while (my ($key, $value) = each %$db) {
1771 print "$key: $value\n";
1774 Please note that when using C<each()>, you should always pass a direct
1775 hash reference, not a lookup. Meaning, you should B<never> do this:
1778 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1780 This causes an infinite loop, because for each iteration, Perl is calling
1781 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1782 it effectively keeps returning the first key over and over again. Instead,
1783 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1787 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1788 reference. This includes inserting, removing and manipulating elements,
1789 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1790 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1791 or simply be a nested array reference inside a hash. Example:
1793 my $db = DBM::Deep->new(
1794 file => "foo-array.db",
1795 type => DBM::Deep->TYPE_ARRAY
1799 push @$db, "bar", "baz";
1800 unshift @$db, "bah";
1802 my $last_elem = pop @$db; # baz
1803 my $first_elem = shift @$db; # bah
1804 my $second_elem = $db->[1]; # bar
1806 my $num_elements = scalar @$db;
1810 In addition to the I<tie()> interface, you can also use a standard OO interface
1811 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1812 array) has its own methods, but both types share the following common methods:
1813 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1819 Stores a new hash key/value pair, or sets an array element value. Takes two
1820 arguments, the hash key or array index, and the new value. The value can be
1821 a scalar, hash ref or array ref. Returns true on success, false on failure.
1823 $db->put("foo", "bar"); # for hashes
1824 $db->put(1, "bar"); # for arrays
1828 Fetches the value of a hash key or array element. Takes one argument: the hash
1829 key or array index. Returns a scalar, hash ref or array ref, depending on the
1832 my $value = $db->get("foo"); # for hashes
1833 my $value = $db->get(1); # for arrays
1837 Checks if a hash key or array index exists. Takes one argument: the hash key
1838 or array index. Returns true if it exists, false if not.
1840 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1841 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1845 Deletes one hash key/value pair or array element. Takes one argument: the hash
1846 key or array index. Returns true on success, false if not found. For arrays,
1847 the remaining elements located after the deleted element are NOT moved over.
1848 The deleted element is essentially just undefined, which is exactly how Perl's
1849 internal arrays work. Please note that the space occupied by the deleted
1850 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1851 below for details and workarounds.
1853 $db->delete("foo"); # for hashes
1854 $db->delete(1); # for arrays
1858 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1859 value. Please note that the space occupied by the deleted keys/values or
1860 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1861 details and workarounds.
1863 $db->clear(); # hashes or arrays
1869 For hashes, DBM::Deep supports all the common methods described above, and the
1870 following additional methods: C<first_key()> and C<next_key()>.
1876 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1877 fetched in an undefined order (which appears random). Takes no arguments,
1878 returns the key as a scalar value.
1880 my $key = $db->first_key();
1884 Returns the "next" key in the hash, given the previous one as the sole argument.
1885 Returns undef if there are no more keys to be fetched.
1887 $key = $db->next_key($key);
1891 Here are some examples of using hashes:
1893 my $db = DBM::Deep->new( "foo.db" );
1895 $db->put("foo", "bar");
1896 print "foo: " . $db->get("foo") . "\n";
1898 $db->put("baz", {}); # new child hash ref
1899 $db->get("baz")->put("buz", "biz");
1900 print "buz: " . $db->get("baz")->get("buz") . "\n";
1902 my $key = $db->first_key();
1904 print "$key: " . $db->get($key) . "\n";
1905 $key = $db->next_key($key);
1908 if ($db->exists("foo")) { $db->delete("foo"); }
1912 For arrays, DBM::Deep supports all the common methods described above, and the
1913 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1914 C<unshift()> and C<splice()>.
1920 Returns the number of elements in the array. Takes no arguments.
1922 my $len = $db->length();
1926 Adds one or more elements onto the end of the array. Accepts scalars, hash
1927 refs or array refs. No return value.
1929 $db->push("foo", "bar", {});
1933 Fetches the last element in the array, and deletes it. Takes no arguments.
1934 Returns undef if array is empty. Returns the element value.
1936 my $elem = $db->pop();
1940 Fetches the first element in the array, deletes it, then shifts all the
1941 remaining elements over to take up the space. Returns the element value. This
1942 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1945 my $elem = $db->shift();
1949 Inserts one or more elements onto the beginning of the array, shifting all
1950 existing elements over to make room. Accepts scalars, hash refs or array refs.
1951 No return value. This method is not recommended with large arrays -- see
1952 <LARGE ARRAYS> below for details.
1954 $db->unshift("foo", "bar", {});
1958 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1959 -f splice> for usage -- it is too complicated to document here. This method is
1960 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1964 Here are some examples of using arrays:
1966 my $db = DBM::Deep->new(
1968 type => DBM::Deep->TYPE_ARRAY
1971 $db->push("bar", "baz");
1972 $db->unshift("foo");
1975 my $len = $db->length();
1976 print "length: $len\n"; # 4
1978 for (my $k=0; $k<$len; $k++) {
1979 print "$k: " . $db->get($k) . "\n";
1982 $db->splice(1, 2, "biz", "baf");
1984 while (my $elem = shift @$db) {
1985 print "shifted: $elem\n";
1990 Enable automatic file locking by passing a true value to the C<locking>
1991 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1993 my $db = DBM::Deep->new(
1998 This causes DBM::Deep to C<flock()> the underlying FileHandle object with exclusive
1999 mode for writes, and shared mode for reads. This is required if you have
2000 multiple processes accessing the same database file, to avoid file corruption.
2001 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
2002 NFS> below for more.
2004 =head2 EXPLICIT LOCKING
2006 You can explicitly lock a database, so it remains locked for multiple
2007 transactions. This is done by calling the C<lock()> method, and passing an
2008 optional lock mode argument (defaults to exclusive mode). This is particularly
2009 useful for things like counters, where the current value needs to be fetched,
2010 then incremented, then stored again.
2013 my $counter = $db->get("counter");
2015 $db->put("counter", $counter);
2024 You can pass C<lock()> an optional argument, which specifies which mode to use
2025 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
2026 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
2027 same as the constants defined in Perl's C<Fcntl> module.
2029 $db->lock( DBM::Deep->LOCK_SH );
2033 If you want to implement your own file locking scheme, be sure to create your
2034 DBM::Deep objects setting the C<volatile> option to true. This hints to DBM::Deep
2035 that the DB file may change between transactions. See L<LOW-LEVEL ACCESS>
2038 =head1 IMPORTING/EXPORTING
2040 You can import existing complex structures by calling the C<import()> method,
2041 and export an entire database into an in-memory structure using the C<export()>
2042 method. Both are examined here.
2046 Say you have an existing hash with nested hashes/arrays inside it. Instead of
2047 walking the structure and adding keys/elements to the database as you go,
2048 simply pass a reference to the C<import()> method. This recursively adds
2049 everything to an existing DBM::Deep object for you. Here is an example:
2054 array1 => [ "elem0", "elem1", "elem2" ],
2056 subkey1 => "subvalue1",
2057 subkey2 => "subvalue2"
2061 my $db = DBM::Deep->new( "foo.db" );
2062 $db->import( $struct );
2064 print $db->{key1} . "\n"; # prints "value1"
2066 This recursively imports the entire C<$struct> object into C<$db>, including
2067 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
2068 keys are merged with the existing ones, replacing if they already exist.
2069 The C<import()> method can be called on any database level (not just the base
2070 level), and works with both hash and array DB types.
2074 B<Note:> Make sure your existing structure has no circular references in it.
2075 These will cause an infinite loop when importing.
2079 Calling the C<export()> method on an existing DBM::Deep object will return
2080 a reference to a new in-memory copy of the database. The export is done
2081 recursively, so all nested hashes/arrays are all exported to standard Perl
2082 objects. Here is an example:
2084 my $db = DBM::Deep->new( "foo.db" );
2086 $db->{key1} = "value1";
2087 $db->{key2} = "value2";
2089 $db->{hash1}->{subkey1} = "subvalue1";
2090 $db->{hash1}->{subkey2} = "subvalue2";
2092 my $struct = $db->export();
2094 print $struct->{key1} . "\n"; # prints "value1"
2096 This makes a complete copy of the database in memory, and returns a reference
2097 to it. The C<export()> method can be called on any database level (not just
2098 the base level), and works with both hash and array DB types. Be careful of
2099 large databases -- you can store a lot more data in a DBM::Deep object than an
2100 in-memory Perl structure.
2104 B<Note:> Make sure your database has no circular references in it.
2105 These will cause an infinite loop when exporting.
2109 DBM::Deep has a number of hooks where you can specify your own Perl function
2110 to perform filtering on incoming or outgoing data. This is a perfect
2111 way to extend the engine, and implement things like real-time compression or
2112 encryption. Filtering applies to the base DB level, and all child hashes /
2113 arrays. Filter hooks can be specified when your DBM::Deep object is first
2114 constructed, or by calling the C<set_filter()> method at any time. There are
2115 four available filter hooks, described below:
2119 =item * filter_store_key
2121 This filter is called whenever a hash key is stored. It
2122 is passed the incoming key, and expected to return a transformed key.
2124 =item * filter_store_value
2126 This filter is called whenever a hash key or array element is stored. It
2127 is passed the incoming value, and expected to return a transformed value.
2129 =item * filter_fetch_key
2131 This filter is called whenever a hash key is fetched (i.e. via
2132 C<first_key()> or C<next_key()>). It is passed the transformed key,
2133 and expected to return the plain key.
2135 =item * filter_fetch_value
2137 This filter is called whenever a hash key or array element is fetched.
2138 It is passed the transformed value, and expected to return the plain value.
2142 Here are the two ways to setup a filter hook:
2144 my $db = DBM::Deep->new(
2146 filter_store_value => \&my_filter_store,
2147 filter_fetch_value => \&my_filter_fetch
2152 $db->set_filter( "filter_store_value", \&my_filter_store );
2153 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
2155 Your filter function will be called only when dealing with SCALAR keys or
2156 values. When nested hashes and arrays are being stored/fetched, filtering
2157 is bypassed. Filters are called as static functions, passed a single SCALAR
2158 argument, and expected to return a single SCALAR value. If you want to
2159 remove a filter, set the function reference to C<undef>:
2161 $db->set_filter( "filter_store_value", undef );
2163 =head2 REAL-TIME ENCRYPTION EXAMPLE
2165 Here is a working example that uses the I<Crypt::Blowfish> module to
2166 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
2167 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
2168 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
2171 use Crypt::Blowfish;
2174 my $cipher = Crypt::CBC->new({
2175 'key' => 'my secret key',
2176 'cipher' => 'Blowfish',
2178 'regenerate_key' => 0,
2179 'padding' => 'space',
2183 my $db = DBM::Deep->new(
2184 file => "foo-encrypt.db",
2185 filter_store_key => \&my_encrypt,
2186 filter_store_value => \&my_encrypt,
2187 filter_fetch_key => \&my_decrypt,
2188 filter_fetch_value => \&my_decrypt,
2191 $db->{key1} = "value1";
2192 $db->{key2} = "value2";
2193 print "key1: " . $db->{key1} . "\n";
2194 print "key2: " . $db->{key2} . "\n";
2200 return $cipher->encrypt( $_[0] );
2203 return $cipher->decrypt( $_[0] );
2206 =head2 REAL-TIME COMPRESSION EXAMPLE
2208 Here is a working example that uses the I<Compress::Zlib> module to do real-time
2209 compression / decompression of keys & values with DBM::Deep Filters.
2210 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
2211 more on I<Compress::Zlib>.
2216 my $db = DBM::Deep->new(
2217 file => "foo-compress.db",
2218 filter_store_key => \&my_compress,
2219 filter_store_value => \&my_compress,
2220 filter_fetch_key => \&my_decompress,
2221 filter_fetch_value => \&my_decompress,
2224 $db->{key1} = "value1";
2225 $db->{key2} = "value2";
2226 print "key1: " . $db->{key1} . "\n";
2227 print "key2: " . $db->{key2} . "\n";
2233 return Compress::Zlib::memGzip( $_[0] ) ;
2236 return Compress::Zlib::memGunzip( $_[0] ) ;
2239 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
2240 actually numerical index numbers, and are not filtered.
2242 =head1 ERROR HANDLING
2244 Most DBM::Deep methods return a true value for success, and call die() on
2245 failure. You can wrap calls in an eval block to catch the die. Also, the
2246 actual error message is stored in an internal scalar, which can be fetched by
2247 calling the C<error()> method.
2249 my $db = DBM::Deep->new( "foo.db" ); # create hash
2250 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
2252 print $db->error(); # prints error message
2254 You can then call C<clear_error()> to clear the current error state.
2258 If you set the C<debug> option to true when creating your DBM::Deep object,
2259 all errors are considered NON-FATAL, and dumped to STDERR. This is only
2260 for debugging purposes.
2262 =head1 LARGEFILE SUPPORT
2264 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
2265 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
2266 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
2267 by calling the static C<set_pack()> method before you do anything else.
2269 DBM::Deep::set_pack(8, 'Q');
2271 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
2272 instead of 32-bit longs. After setting these values your DB files have a
2273 theoretical maximum size of 16 XB (exabytes).
2277 B<Note:> Changing these values will B<NOT> work for existing database files.
2278 Only change this for new files, and make sure it stays set consistently
2279 throughout the file's life. If you do set these values, you can no longer
2280 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
2281 back to 32-bit mode.
2285 B<Note:> I have not personally tested files > 2 GB -- all my systems have
2286 only a 32-bit Perl. However, I have received user reports that this does
2289 =head1 LOW-LEVEL ACCESS
2291 If you require low-level access to the underlying FileHandle that DBM::Deep uses,
2292 you can call the C<fh()> method, which returns the handle:
2296 This method can be called on the root level of the datbase, or any child
2297 hashes or arrays. All levels share a I<root> structure, which contains things
2298 like the FileHandle, a reference counter, and all your options you specified
2299 when you created the object. You can get access to this root structure by
2300 calling the C<root()> method.
2302 my $root = $db->root();
2304 This is useful for changing options after the object has already been created,
2305 such as enabling/disabling locking, volatile or debug modes. You can also
2306 store your own temporary user data in this structure (be wary of name
2307 collision), which is then accessible from any child hash or array.
2309 =head1 CUSTOM DIGEST ALGORITHM
2311 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
2312 keys. However you can override this, and use another algorithm (such as SHA-256)
2313 or even write your own. But please note that DBM::Deep currently expects zero
2314 collisions, so your algorithm has to be I<perfect>, so to speak.
2315 Collision detection may be introduced in a later version.
2319 You can specify a custom digest algorithm by calling the static C<set_digest()>
2320 function, passing a reference to a subroutine, and the length of the algorithm's
2321 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
2322 objects. Here is a working example that uses a 256-bit hash from the
2323 I<Digest::SHA256> module. Please see
2324 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
2329 my $context = Digest::SHA256::new(256);
2331 DBM::Deep::set_digest( \&my_digest, 32 );
2333 my $db = DBM::Deep->new( "foo-sha.db" );
2335 $db->{key1} = "value1";
2336 $db->{key2} = "value2";
2337 print "key1: " . $db->{key1} . "\n";
2338 print "key2: " . $db->{key2} . "\n";
2344 return substr( $context->hash($_[0]), 0, 32 );
2347 B<Note:> Your returned digest strings must be B<EXACTLY> the number
2348 of bytes you specify in the C<set_digest()> function (in this case 32).
2350 =head1 CIRCULAR REFERENCES
2352 DBM::Deep has B<experimental> support for circular references. Meaning you
2353 can have a nested hash key or array element that points to a parent object.
2354 This relationship is stored in the DB file, and is preserved between sessions.
2357 my $db = DBM::Deep->new( "foo.db" );
2360 $db->{circle} = $db; # ref to self
2362 print $db->{foo} . "\n"; # prints "foo"
2363 print $db->{circle}->{foo} . "\n"; # prints "foo" again
2365 One catch is, passing the object to a function that recursively walks the
2366 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
2367 C<export()> methods) will result in an infinite loop. The other catch is,
2368 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
2369 or C<next_key()> methods), you will get the I<target object's key>, not the
2370 ref's key. This gets even more interesting with the above example, where
2371 the I<circle> key points to the base DB object, which technically doesn't
2372 have a key. So I made DBM::Deep return "[base]" as the key name in that
2375 =head1 CAVEATS / ISSUES / BUGS
2377 This section describes all the known issues with DBM::Deep. It you have found
2378 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
2380 =head2 UNUSED SPACE RECOVERY
2382 One major caveat with DBM::Deep is that space occupied by existing keys and
2383 values is not recovered when they are deleted. Meaning if you keep deleting
2384 and adding new keys, your file will continuously grow. I am working on this,
2385 but in the meantime you can call the built-in C<optimize()> method from time to
2386 time (perhaps in a crontab or something) to recover all your unused space.
2388 $db->optimize(); # returns true on success
2390 This rebuilds the ENTIRE database into a new file, then moves it on top of
2391 the original. The new file will have no unused space, thus it will take up as
2392 little disk space as possible. Please note that this operation can take
2393 a long time for large files, and you need enough disk space to temporarily hold
2394 2 copies of your DB file. The temporary file is created in the same directory
2395 as the original, named with a ".tmp" extension, and is deleted when the
2396 operation completes. Oh, and if locking is enabled, the DB is automatically
2397 locked for the entire duration of the copy.
2401 B<WARNING:> Only call optimize() on the top-level node of the database, and
2402 make sure there are no child references lying around. DBM::Deep keeps a reference
2403 counter, and if it is greater than 1, optimize() will abort and return undef.
2405 =head2 AUTOVIVIFICATION
2407 Unfortunately, autovivification doesn't work with tied hashes. This appears to
2408 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
2409 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
2410 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
2411 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
2414 $db->{foo}->{bar} = "hello";
2416 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
2417 being an empty hash. Try this instead, which works fine:
2419 $db->{foo} = { bar => "hello" };
2421 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
2422 the execution path, and Perl indeed passes an empty hash to the STORE() method.
2423 Probably a bug in Perl.
2425 =head2 FILE CORRUPTION
2427 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2428 for a 32-bit signature when opened, but other corruption in files can cause
2429 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2430 stuck in an infinite loop depending on the level of corruption. File write
2431 operations are not checked for failure (for speed), so if you happen to run
2432 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2433 be addressed in a later version of DBM::Deep.
2437 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2438 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2439 about setting up your NFS server with a locking daemon, then using lockf() to
2440 lock your files, but your milage may vary there as well. From what I
2441 understand, there is no real way to do it. However, if you need access to the
2442 underlying FileHandle in DBM::Deep for using some other kind of locking scheme like
2443 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2445 =head2 COPYING OBJECTS
2447 Beware of copying tied objects in Perl. Very strange things can happen.
2448 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2449 returns a new, blessed, tied hash or array to the same level in the DB.
2451 my $copy = $db->clone();
2455 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2456 These functions cause every element in the array to move, which can be murder
2457 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2458 a different location. This may be addressed in a later version.
2462 This section discusses DBM::Deep's speed and memory usage.
2466 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2467 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2468 multi-level hash/array support, and cross-platform FTPable files. Even so,
2469 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2470 with huge databases. Here is some test data:
2472 Adding 1,000,000 keys to new DB file...
2474 At 100 keys, avg. speed is 2,703 keys/sec
2475 At 200 keys, avg. speed is 2,642 keys/sec
2476 At 300 keys, avg. speed is 2,598 keys/sec
2477 At 400 keys, avg. speed is 2,578 keys/sec
2478 At 500 keys, avg. speed is 2,722 keys/sec
2479 At 600 keys, avg. speed is 2,628 keys/sec
2480 At 700 keys, avg. speed is 2,700 keys/sec
2481 At 800 keys, avg. speed is 2,607 keys/sec
2482 At 900 keys, avg. speed is 2,190 keys/sec
2483 At 1,000 keys, avg. speed is 2,570 keys/sec
2484 At 2,000 keys, avg. speed is 2,417 keys/sec
2485 At 3,000 keys, avg. speed is 1,982 keys/sec
2486 At 4,000 keys, avg. speed is 1,568 keys/sec
2487 At 5,000 keys, avg. speed is 1,533 keys/sec
2488 At 6,000 keys, avg. speed is 1,787 keys/sec
2489 At 7,000 keys, avg. speed is 1,977 keys/sec
2490 At 8,000 keys, avg. speed is 2,028 keys/sec
2491 At 9,000 keys, avg. speed is 2,077 keys/sec
2492 At 10,000 keys, avg. speed is 2,031 keys/sec
2493 At 20,000 keys, avg. speed is 1,970 keys/sec
2494 At 30,000 keys, avg. speed is 2,050 keys/sec
2495 At 40,000 keys, avg. speed is 2,073 keys/sec
2496 At 50,000 keys, avg. speed is 1,973 keys/sec
2497 At 60,000 keys, avg. speed is 1,914 keys/sec
2498 At 70,000 keys, avg. speed is 2,091 keys/sec
2499 At 80,000 keys, avg. speed is 2,103 keys/sec
2500 At 90,000 keys, avg. speed is 1,886 keys/sec
2501 At 100,000 keys, avg. speed is 1,970 keys/sec
2502 At 200,000 keys, avg. speed is 2,053 keys/sec
2503 At 300,000 keys, avg. speed is 1,697 keys/sec
2504 At 400,000 keys, avg. speed is 1,838 keys/sec
2505 At 500,000 keys, avg. speed is 1,941 keys/sec
2506 At 600,000 keys, avg. speed is 1,930 keys/sec
2507 At 700,000 keys, avg. speed is 1,735 keys/sec
2508 At 800,000 keys, avg. speed is 1,795 keys/sec
2509 At 900,000 keys, avg. speed is 1,221 keys/sec
2510 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2512 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2513 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2514 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2515 Run time was 12 min 3 sec.
2519 One of the great things about DBM::Deep is that it uses very little memory.
2520 Even with huge databases (1,000,000+ keys) you will not see much increased
2521 memory on your process. DBM::Deep relies solely on the filesystem for storing
2522 and fetching data. Here is output from I</usr/bin/top> before even opening a
2525 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2526 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2528 Basically the process is taking 2,716K of memory. And here is the same
2529 process after storing and fetching 1,000,000 keys:
2531 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2532 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2534 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2535 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2537 =head1 DB FILE FORMAT
2539 In case you were interested in the underlying DB file format, it is documented
2540 here in this section. You don't need to know this to use the module, it's just
2541 included for reference.
2545 DBM::Deep files always start with a 32-bit signature to identify the file type.
2546 This is at offset 0. The signature is "DPDB" in network byte order. This is
2547 checked when the file is opened.
2551 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2552 has a standard header containing the type of data, the length of data, and then
2553 the data itself. The type is a single character (1 byte), the length is a
2554 32-bit unsigned long in network byte order, and the data is, well, the data.
2555 Here is how it unfolds:
2559 Immediately after the 32-bit file signature is the I<Master Index> record.
2560 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2561 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2562 depending on how the DBM::Deep object was constructed.
2566 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2567 number). The first 8-bit char of the MD5 signature is the offset into the
2568 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2569 index element is a file offset of the next tag for the key/element in question,
2570 which is usually a I<Bucket List> tag (see below).
2574 The next tag I<could> be another index, depending on how many keys/elements
2575 exist. See L<RE-INDEXING> below for details.
2579 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2580 file offsets to where the actual data is stored. It starts with a standard
2581 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2582 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2583 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2584 When the list fills up, a I<Re-Index> operation is performed (See
2585 L<RE-INDEXING> below).
2589 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2590 index/value pair (in array mode). It starts with a standard tag header with
2591 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2592 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2593 header. The size reported in the tag header is only for the value, but then,
2594 just after the value is another size (32-bit unsigned long) and then the plain
2595 key itself. Since the value is likely to be fetched more often than the plain
2596 key, I figured it would be I<slightly> faster to store the value first.
2600 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2601 record for the nested structure, where the process begins all over again.
2605 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2606 exhausted. Then, when another key/element comes in, the list is converted to a
2607 new index record. However, this index will look at the next char in the MD5
2608 hash, and arrange new Bucket List pointers accordingly. This process is called
2609 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2610 17 (16 + new one) keys/elements are removed from the old Bucket List and
2611 inserted into the new index. Several new Bucket Lists are created in the
2612 process, as a new MD5 char from the key is being examined (it is unlikely that
2613 the keys will all share the same next char of their MD5s).
2617 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2618 when the Bucket Lists will turn into indexes, but the first round tends to
2619 happen right around 4,000 keys. You will see a I<slight> decrease in
2620 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2621 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2622 right around 900,000 keys. This process can continue nearly indefinitely --
2623 right up until the point the I<MD5> signatures start colliding with each other,
2624 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2625 getting struck by lightning while you are walking to cash in your tickets.
2626 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2627 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2628 this is 340 unodecillion, but don't quote me).
2632 When a new key/element is stored, the key (or index number) is first ran through
2633 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2634 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2635 for the first char of the signature (in this case I<b>). If it does not exist,
2636 a new I<Bucket List> is created for our key (and the next 15 future keys that
2637 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2638 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2639 this point, unless we are replacing an existing I<Bucket>), where the actual
2640 data will be stored.
2644 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2645 (or index number), then walking along the indexes. If there are enough
2646 keys/elements in this DB level, there might be nested indexes, each linked to
2647 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2648 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2649 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2650 plain key are stored.
2654 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2655 methods. In this process the indexes are walked systematically, and each key
2656 fetched in increasing MD5 order (which is why it appears random). Once the
2657 I<Bucket> is found, the value is skipped the plain key returned instead.
2658 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2659 alphabetically sorted. This only happens on an index-level -- as soon as the
2660 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2661 so it's pretty much undefined how the keys will come out -- just like Perl's
2664 =head1 CODE COVERAGE
2666 I use B<Devel::Cover> to test the code coverage of my tests, below is the B<Devel::Cover> report on this
2667 module's test suite.
2669 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2670 File stmt bran cond sub pod time total
2671 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2672 blib/lib/DBM/Deep.pm 94.1 82.9 74.5 98.0 10.5 98.1 88.2
2673 blib/lib/DBM/Deep/Array.pm 97.8 83.3 50.0 100.0 n/a 1.6 94.4
2674 blib/lib/DBM/Deep/Hash.pm 93.3 85.7 100.0 100.0 n/a 0.3 92.7
2675 Total 94.5 83.1 75.5 98.4 10.5 100.0 89.0
2676 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2680 Joseph Huckaby, L<jhuckaby@cpan.org>
2682 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2686 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2687 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2691 Copyright (c) 2002-2005 Joseph Huckaby. All Rights Reserved.
2692 This is free software, you may use it and distribute it under the
2693 same terms as Perl itself.