diff options
Diffstat (limited to 'Master/tlpkg/tlperl/lib/DBM/Deep')
| -rwxr-xr-x | Master/tlpkg/tlperl/lib/DBM/Deep/Array.pm | 414 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl/lib/DBM/Deep/Cookbook.pod | 199 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl/lib/DBM/Deep/Engine.pm | 2325 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl/lib/DBM/Deep/File.pm | 277 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl/lib/DBM/Deep/Hash.pm | 136 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl/lib/DBM/Deep/Internals.pod | 281 |
6 files changed, 3632 insertions, 0 deletions
diff --git a/Master/tlpkg/tlperl/lib/DBM/Deep/Array.pm b/Master/tlpkg/tlperl/lib/DBM/Deep/Array.pm new file mode 100755 index 00000000000..38c51869fda --- /dev/null +++ b/Master/tlpkg/tlperl/lib/DBM/Deep/Array.pm @@ -0,0 +1,414 @@ +package DBM::Deep::Array; + +use 5.006_000; + +use strict; +use warnings; + +our $VERSION = q(1.0013); + +# This is to allow DBM::Deep::Array to handle negative indices on +# its own. Otherwise, Perl would intercept the call to negative +# indices for us. This was causing bugs for negative index handling. +our $NEGATIVE_INDICES = 1; + +use base 'DBM::Deep'; + +use Scalar::Util (); + +sub _get_self { + eval { local $SIG{'__DIE__'}; tied( @{$_[0]} ) } || $_[0] +} + +sub _repr { [] } + +sub TIEARRAY { + my $class = shift; + my $args = $class->_get_args( @_ ); + + $args->{type} = $class->TYPE_ARRAY; + + return $class->_init($args); +} + +sub FETCH { + my $self = shift->_get_self; + my ($key) = @_; + + $self->lock( $self->LOCK_SH ); + + if ( !defined $key ) { + $self->unlock; + DBM::Deep->_throw_error( "Cannot use an undefined array index." ); + } + elsif ( $key =~ /^-?\d+$/ ) { + if ( $key < 0 ) { + $key += $self->FETCHSIZE; + unless ( $key >= 0 ) { + $self->unlock; + return; + } + } + } + elsif ( $key ne 'length' ) { + $self->unlock; + DBM::Deep->_throw_error( "Cannot use '$key' as an array index." ); + } + + my $rv = $self->SUPER::FETCH( $key ); + + $self->unlock; + + return $rv; +} + +sub STORE { + my $self = shift->_get_self; + my ($key, $value) = @_; + + $self->lock( $self->LOCK_EX ); + + my $size; + my $idx_is_numeric; + if ( !defined $key ) { + $self->unlock; + DBM::Deep->_throw_error( "Cannot use an undefined array index." ); + } + elsif ( $key =~ /^-?\d+$/ ) { + $idx_is_numeric = 1; + if ( $key < 0 ) { + $size = $self->FETCHSIZE; + if ( $key + $size < 0 ) { + die( "Modification of non-creatable array value attempted, subscript $key" ); + } + $key += $size + } + } + elsif ( $key ne 'length' ) { + $self->unlock; + DBM::Deep->_throw_error( "Cannot use '$key' as an array index." ); + } + + my $rv = $self->SUPER::STORE( $key, $value ); + + if ( $idx_is_numeric ) { + $size = $self->FETCHSIZE unless defined $size; + if ( $key >= $size ) { + $self->STORESIZE( $key + 1 ); + } + } + + $self->unlock; + + return $rv; +} + +sub EXISTS { + my $self = shift->_get_self; + my ($key) = @_; + + $self->lock( $self->LOCK_SH ); + + if ( !defined $key ) { + $self->unlock; + DBM::Deep->_throw_error( "Cannot use an undefined array index." ); + } + elsif ( $key =~ /^-?\d+$/ ) { + if ( $key < 0 ) { + $key += $self->FETCHSIZE; + unless ( $key >= 0 ) { + $self->unlock; + return; + } + } + } + elsif ( $key ne 'length' ) { + $self->unlock; + DBM::Deep->_throw_error( "Cannot use '$key' as an array index." ); + } + + my $rv = $self->SUPER::EXISTS( $key ); + + $self->unlock; + + return $rv; +} + +sub DELETE { + my $self = shift->_get_self; + my ($key) = @_; + warn "ARRAY::DELETE($self,$key)\n" if DBM::Deep::DEBUG; + + $self->lock( $self->LOCK_EX ); + + my $size = $self->FETCHSIZE; + if ( !defined $key ) { + $self->unlock; + DBM::Deep->_throw_error( "Cannot use an undefined array index." ); + } + elsif ( $key =~ /^-?\d+$/ ) { + if ( $key < 0 ) { + $key += $size; + unless ( $key >= 0 ) { + $self->unlock; + return; + } + } + } + elsif ( $key ne 'length' ) { + $self->unlock; + DBM::Deep->_throw_error( "Cannot use '$key' as an array index." ); + } + + my $rv = $self->SUPER::DELETE( $key ); + + if ($rv && $key == $size - 1) { + $self->STORESIZE( $key ); + } + + $self->unlock; + + return $rv; +} + +# Now that we have a real Reference sector, we should store arrayzize there. However, +# arraysize needs to be transactionally-aware, so a simple location to store it isn't +# going to work. +sub FETCHSIZE { + my $self = shift->_get_self; + + $self->lock( $self->LOCK_SH ); + + my $SAVE_FILTER = $self->_storage->{filter_fetch_value}; + $self->_storage->{filter_fetch_value} = undef; + + my $size = $self->FETCH('length') || 0; + + $self->_storage->{filter_fetch_value} = $SAVE_FILTER; + + $self->unlock; + + return $size; +} + +sub STORESIZE { + my $self = shift->_get_self; + my ($new_length) = @_; + + $self->lock( $self->LOCK_EX ); + + my $SAVE_FILTER = $self->_storage->{filter_store_value}; + $self->_storage->{filter_store_value} = undef; + + my $result = $self->STORE('length', $new_length, 'length'); + + $self->_storage->{filter_store_value} = $SAVE_FILTER; + + $self->unlock; + + return $result; +} + +sub POP { + my $self = shift->_get_self; + + $self->lock( $self->LOCK_EX ); + + my $length = $self->FETCHSIZE(); + + if ($length) { + my $content = $self->FETCH( $length - 1 ); + $self->DELETE( $length - 1 ); + + $self->unlock; + + return $content; + } + else { + $self->unlock; + return; + } +} + +sub PUSH { + my $self = shift->_get_self; + + $self->lock( $self->LOCK_EX ); + + my $length = $self->FETCHSIZE(); + + while (my $content = shift @_) { + $self->STORE( $length, $content ); + $length++; + } + + $self->unlock; + + return $length; +} + +# XXX This really needs to be something more direct within the file, not a +# fetch and re-store. -RobK, 2007-09-20 +sub _move_value { + my $self = shift; + my ($old_key, $new_key) = @_; + + return $self->_engine->make_reference( $self, $old_key, $new_key ); +} + +sub SHIFT { + my $self = shift->_get_self; + warn "SHIFT($self)\n" if DBM::Deep::DEBUG; + + $self->lock( $self->LOCK_EX ); + + my $length = $self->FETCHSIZE(); + + if ( !$length ) { + $self->unlock; + return; + } + + my $content = $self->DELETE( 0 ); + + # Unless the deletion above has cleared the array ... + if ( $length > 1 ) { + for (my $i = 0; $i < $length - 1; $i++) { + $self->_move_value( $i+1, $i ); + } + + $self->DELETE( $length - 1 ); + } + + $self->unlock; + + return $content; +} + +sub UNSHIFT { + my $self = shift->_get_self; + my @new_elements = @_; + + $self->lock( $self->LOCK_EX ); + + my $length = $self->FETCHSIZE(); + my $new_size = scalar @new_elements; + + if ($length) { + for (my $i = $length - 1; $i >= 0; $i--) { + $self->_move_value( $i, $i+$new_size ); + } + + $self->STORESIZE( $length + $new_size ); + } + + for (my $i = 0; $i < $new_size; $i++) { + $self->STORE( $i, $new_elements[$i] ); + } + + $self->unlock; + + return $length + $new_size; +} + +sub SPLICE { + my $self = shift->_get_self; + + $self->lock( $self->LOCK_EX ); + + my $length = $self->FETCHSIZE(); + + ## + # Calculate offset and length of splice + ## + my $offset = shift; + $offset = 0 unless defined $offset; + if ($offset < 0) { $offset += $length; } + + my $splice_length; + if (scalar @_) { $splice_length = shift; } + else { $splice_length = $length - $offset; } + if ($splice_length < 0) { $splice_length += ($length - $offset); } + + ## + # Setup array with new elements, and copy out old elements for return + ## + my @new_elements = @_; + my $new_size = scalar @new_elements; + + my @old_elements = map { + $self->FETCH( $_ ) + } $offset .. ($offset + $splice_length - 1); + + ## + # Adjust array length, and shift elements to accomodate new section. + ## + if ( $new_size != $splice_length ) { + if ($new_size > $splice_length) { + for (my $i = $length - 1; $i >= $offset + $splice_length; $i--) { + $self->_move_value( $i, $i + ($new_size - $splice_length) ); + } + $self->STORESIZE( $length + $new_size - $splice_length ); + } + else { + for (my $i = $offset + $splice_length; $i < $length; $i++) { + $self->_move_value( $i, $i + ($new_size - $splice_length) ); + } + for (my $i = 0; $i < $splice_length - $new_size; $i++) { + $self->DELETE( $length - 1 ); + $length--; + } + } + } + + ## + # Insert new elements into array + ## + for (my $i = $offset; $i < $offset + $new_size; $i++) { + $self->STORE( $i, shift @new_elements ); + } + + $self->unlock; + + ## + # Return deleted section, or last element in scalar context. + ## + return wantarray ? @old_elements : $old_elements[-1]; +} + +# We don't need to populate it, yet. +# It will be useful, though, when we split out HASH and ARRAY +sub EXTEND { + ## + # Perl will call EXTEND() when the array is likely to grow. + # We don't care, but include it because it gets called at times. + ## +} + +sub _copy_node { + my $self = shift; + my ($db_temp) = @_; + + my $length = $self->length(); + for (my $index = 0; $index < $length; $index++) { + $self->_copy_value( \$db_temp->[$index], $self->get($index) ); + } + + return 1; +} + +## +# Public method aliases +## +sub length { (shift)->FETCHSIZE(@_) } +sub pop { (shift)->POP(@_) } +sub push { (shift)->PUSH(@_) } +sub unshift { (shift)->UNSHIFT(@_) } +sub splice { (shift)->SPLICE(@_) } + +# This must be last otherwise we have to qualify all other calls to shift +# as calls to CORE::shift +sub shift { (CORE::shift)->SHIFT(@_) } + +1; +__END__ diff --git a/Master/tlpkg/tlperl/lib/DBM/Deep/Cookbook.pod b/Master/tlpkg/tlperl/lib/DBM/Deep/Cookbook.pod new file mode 100755 index 00000000000..4ca50b97a39 --- /dev/null +++ b/Master/tlpkg/tlperl/lib/DBM/Deep/Cookbook.pod @@ -0,0 +1,199 @@ +=head1 NAME + +DBM::Deep::Cookbook + +=head1 DESCRIPTION + +This is the Cookbook for L<DBM::Deep>. It contains useful tips and tricks, +plus some examples of how to do common tasks. + +=head1 RECIPES + +=head2 UTF8 data + +When you're using UTF8 data, you may run into the "Wide character in print" +warning. To fix that in 5.8+, do the following: + + my $db = DBM::Deep->new( ... ); + binmode $db->_fh, ":utf8"; + +In 5.6, you will have to do the following: + + my $db = DBM::Deep->new( ... ); + $db->set_filter( 'store_value' => sub { pack "U0C*", unpack "C*", $_[0] } ); + $db->set_filter( 'retrieve_value' => sub { pack "C*", unpack "U0C*", $_[0] } ); + +In a future version, you will be able to specify C<utf8 =E<gt> 1> and +L<DBM::Deep> will do these things for you. + +=head2 Real-time Encryption Example + +B<NOTE>: This is just an example of how to write a filter. This most +definitely should B<NOT> be taken as a proper way to write a filter that does +encryption. + +Here is a working example that uses the I<Crypt::Blowfish> module to +do real-time encryption / decryption of keys & values with DBM::Deep Filters. +Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more +on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module. + + use DBM::Deep; + use Crypt::Blowfish; + use Crypt::CBC; + + my $cipher = Crypt::CBC->new({ + 'key' => 'my secret key', + 'cipher' => 'Blowfish', + 'iv' => '$KJh#(}q', + 'regenerate_key' => 0, + 'padding' => 'space', + 'prepend_iv' => 0 + }); + + my $db = DBM::Deep->new( + file => "foo-encrypt.db", + filter_store_key => \&my_encrypt, + filter_store_value => \&my_encrypt, + filter_fetch_key => \&my_decrypt, + filter_fetch_value => \&my_decrypt, + ); + + $db->{key1} = "value1"; + $db->{key2} = "value2"; + print "key1: " . $db->{key1} . "\n"; + print "key2: " . $db->{key2} . "\n"; + + undef $db; + exit; + + sub my_encrypt { + return $cipher->encrypt( $_[0] ); + } + sub my_decrypt { + return $cipher->decrypt( $_[0] ); + } + +=head2 Real-time Compression Example + +Here is a working example that uses the I<Compress::Zlib> module to do real-time +compression / decompression of keys & values with DBM::Deep Filters. +Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for +more on I<Compress::Zlib>. + + use DBM::Deep; + use Compress::Zlib; + + my $db = DBM::Deep->new( + file => "foo-compress.db", + filter_store_key => \&my_compress, + filter_store_value => \&my_compress, + filter_fetch_key => \&my_decompress, + filter_fetch_value => \&my_decompress, + ); + + $db->{key1} = "value1"; + $db->{key2} = "value2"; + print "key1: " . $db->{key1} . "\n"; + print "key2: " . $db->{key2} . "\n"; + + undef $db; + exit; + + sub my_compress { + return Compress::Zlib::memGzip( $_[0] ) ; + } + sub my_decompress { + return Compress::Zlib::memGunzip( $_[0] ) ; + } + +B<Note:> Filtering of keys only applies to hashes. Array "keys" are +actually numerical index numbers, and are not filtered. + +=head1 Custom Digest Algorithm + +DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing +keys. However you can override this, and use another algorithm (such as SHA-256) +or even write your own. But please note that DBM::Deep currently expects zero +collisions, so your algorithm has to be I<perfect>, so to speak. Collision +detection may be introduced in a later version. + +You can specify a custom digest algorithm by passing it into the parameter +list for new(), passing a reference to a subroutine as the 'digest' parameter, +and the length of the algorithm's hashes (in bytes) as the 'hash_size' +parameter. Here is a working example that uses a 256-bit hash from the +I<Digest::SHA256> module. Please see +L<http://search.cpan.org/search?module=Digest::SHA256> for more information. + + use DBM::Deep; + use Digest::SHA256; + + my $context = Digest::SHA256::new(256); + + my $db = DBM::Deep->new( + filename => "foo-sha.db", + digest => \&my_digest, + hash_size => 32, + ); + + $db->{key1} = "value1"; + $db->{key2} = "value2"; + print "key1: " . $db->{key1} . "\n"; + print "key2: " . $db->{key2} . "\n"; + + undef $db; + exit; + + sub my_digest { + return substr( $context->hash($_[0]), 0, 32 ); + } + +B<Note:> Your returned digest strings must be B<EXACTLY> the number +of bytes you specify in the hash_size parameter (in this case 32). Undefined +behavior will occur otherwise. + +B<Note:> If you do choose to use a custom digest algorithm, you must set it +every time you access this file. Otherwise, the default (MD5) will be used. + +=head1 PERFORMANCE + +Because DBM::Deep is a conncurrent datastore, every change is flushed to disk +immediately and every read goes to disk. This means that DBM::Deep functions +at the speed of disk (generally 10-20ms) vs. the speed of RAM (generally +50-70ns), or at least 150-200x slower than the comparable in-memory +datastructure in Perl. + +There are several techniques you can use to speed up how DBM::Deep functions. + +=over 4 + +=item * Put it on a ramdisk + +The easiest and quickest mechanism to making DBM::Deep run faster is to create +a ramdisk and locate the DBM::Deep file there. Doing this as an option may +become a feature of DBM::Deep, assuming there is a good ramdisk wrapper on CPAN. + +=item * Work at the tightest level possible + +It is much faster to assign the level of your db that you are working with to +an intermediate variable than to re-look it up every time. Thus + + # BAD + while ( my ($k, $v) = each %{$db->{foo}{bar}{baz}} ) { + ... + } + + # GOOD + my $x = $db->{foo}{bar}{baz}; + while ( my ($k, $v) = each %$x ) { + ... + } + +=item * Make your file as tight as possible + +If you know that you are not going to use more than 65K in your database, +consider using the C<pack_size =E<gt> 'small'> option. This will instruct +DBM::Deep to use 16bit addresses, meaning that the seek times will be less. + +=back + +=cut diff --git a/Master/tlpkg/tlperl/lib/DBM/Deep/Engine.pm b/Master/tlpkg/tlperl/lib/DBM/Deep/Engine.pm new file mode 100755 index 00000000000..fa04b4f3239 --- /dev/null +++ b/Master/tlpkg/tlperl/lib/DBM/Deep/Engine.pm @@ -0,0 +1,2325 @@ +package DBM::Deep::Engine; + +use 5.006_000; + +use strict; +use warnings; + +our $VERSION = q(1.0013); + +# Never import symbols into our namespace. We are a class, not a library. +# -RobK, 2008-05-27 +use Scalar::Util (); + +#use Data::Dumper (); + +# File-wide notes: +# * Every method in here assumes that the storage has been appropriately +# safeguarded. This can be anything from flock() to some sort of manual +# mutex. But, it's the caller's responsability to make sure that this has +# been done. + +# Setup file and tag signatures. These should never change. +sub SIG_FILE () { 'DPDB' } +sub SIG_HEADER () { 'h' } +sub SIG_HASH () { 'H' } +sub SIG_ARRAY () { 'A' } +sub SIG_NULL () { 'N' } +sub SIG_DATA () { 'D' } +sub SIG_INDEX () { 'I' } +sub SIG_BLIST () { 'B' } +sub SIG_FREE () { 'F' } +sub SIG_SIZE () { 1 } + +my $STALE_SIZE = 2; + +# Please refer to the pack() documentation for further information +my %StP = ( + 1 => 'C', # Unsigned char value (no order needed as it's just one byte) + 2 => 'n', # Unsigned short in "network" (big-endian) order + 4 => 'N', # Unsigned long in "network" (big-endian) order + 8 => 'Q', # Usigned quad (no order specified, presumably machine-dependent) +); + +################################################################################ + +sub new { + my $class = shift; + my ($args) = @_; + + my $self = bless { + byte_size => 4, + + digest => undef, + hash_size => 16, # In bytes + hash_chars => 256, # Number of chars the algorithm uses per byte + max_buckets => 16, + num_txns => 1, # The HEAD + trans_id => 0, # Default to the HEAD + + data_sector_size => 64, # Size in bytes of each data sector + + entries => {}, # This is the list of entries for transactions + storage => undef, + }, $class; + + # Never allow byte_size to be set directly. + delete $args->{byte_size}; + if ( defined $args->{pack_size} ) { + if ( lc $args->{pack_size} eq 'small' ) { + $args->{byte_size} = 2; + } + elsif ( lc $args->{pack_size} eq 'medium' ) { + $args->{byte_size} = 4; + } + elsif ( lc $args->{pack_size} eq 'large' ) { + $args->{byte_size} = 8; + } + else { + DBM::Deep->_throw_error( "Unknown pack_size value: '$args->{pack_size}'" ); + } + } + + # Grab the parameters we want to use + foreach my $param ( keys %$self ) { + next unless exists $args->{$param}; + $self->{$param} = $args->{$param}; + } + + my %validations = ( + max_buckets => { floor => 16, ceil => 256 }, + num_txns => { floor => 1, ceil => 255 }, + data_sector_size => { floor => 32, ceil => 256 }, + ); + + while ( my ($attr, $c) = each %validations ) { + if ( !defined $self->{$attr} + || !length $self->{$attr} + || $self->{$attr} =~ /\D/ + || $self->{$attr} < $c->{floor} + ) { + $self->{$attr} = '(undef)' if !defined $self->{$attr}; + warn "Floor of $attr is $c->{floor}. Setting it to $c->{floor} from '$self->{$attr}'\n"; + $self->{$attr} = $c->{floor}; + } + elsif ( $self->{$attr} > $c->{ceil} ) { + warn "Ceiling of $attr is $c->{ceil}. Setting it to $c->{ceil} from '$self->{$attr}'\n"; + $self->{$attr} = $c->{ceil}; + } + } + + if ( !$self->{digest} ) { + require Digest::MD5; + $self->{digest} = \&Digest::MD5::md5; + } + + return $self; +} + +################################################################################ + +sub read_value { + my $self = shift; + my ($obj, $key) = @_; + + # This will be a Reference sector + my $sector = $self->_load_sector( $obj->_base_offset ) + or return; + + if ( $sector->staleness != $obj->_staleness ) { + return; + } + + my $key_md5 = $self->_apply_digest( $key ); + + my $value_sector = $sector->get_data_for({ + key_md5 => $key_md5, + allow_head => 1, + }); + + unless ( $value_sector ) { + $value_sector = DBM::Deep::Engine::Sector::Null->new({ + engine => $self, + data => undef, + }); + + $sector->write_data({ + key_md5 => $key_md5, + key => $key, + value => $value_sector, + }); + } + + return $value_sector->data; +} + +sub get_classname { + my $self = shift; + my ($obj) = @_; + + # This will be a Reference sector + my $sector = $self->_load_sector( $obj->_base_offset ) + or DBM::Deep->_throw_error( "How did get_classname fail (no sector for '$obj')?!" ); + + if ( $sector->staleness != $obj->_staleness ) { + return; + } + + return $sector->get_classname; +} + +sub make_reference { + my $self = shift; + my ($obj, $old_key, $new_key) = @_; + + # This will be a Reference sector + my $sector = $self->_load_sector( $obj->_base_offset ) + or DBM::Deep->_throw_error( "How did get_classname fail (no sector for '$obj')?!" ); + + if ( $sector->staleness != $obj->_staleness ) { + return; + } + + my $old_md5 = $self->_apply_digest( $old_key ); + + my $value_sector = $sector->get_data_for({ + key_md5 => $old_md5, + allow_head => 1, + }); + + unless ( $value_sector ) { + $value_sector = DBM::Deep::Engine::Sector::Null->new({ + engine => $self, + data => undef, + }); + + $sector->write_data({ + key_md5 => $old_md5, + key => $old_key, + value => $value_sector, + }); + } + + if ( $value_sector->isa( 'DBM::Deep::Engine::Sector::Reference' ) ) { + $sector->write_data({ + key => $new_key, + key_md5 => $self->_apply_digest( $new_key ), + value => $value_sector, + }); + $value_sector->increment_refcount; + } + else { + $sector->write_data({ + key => $new_key, + key_md5 => $self->_apply_digest( $new_key ), + value => $value_sector->clone, + }); + } +} + +sub key_exists { + my $self = shift; + my ($obj, $key) = @_; + + # This will be a Reference sector + my $sector = $self->_load_sector( $obj->_base_offset ) + or return ''; + + if ( $sector->staleness != $obj->_staleness ) { + return ''; + } + + my $data = $sector->get_data_for({ + key_md5 => $self->_apply_digest( $key ), + allow_head => 1, + }); + + # exists() returns 1 or '' for true/false. + return $data ? 1 : ''; +} + +sub delete_key { + my $self = shift; + my ($obj, $key) = @_; + + my $sector = $self->_load_sector( $obj->_base_offset ) + or return; + + if ( $sector->staleness != $obj->_staleness ) { + return; + } + + return $sector->delete_key({ + key_md5 => $self->_apply_digest( $key ), + allow_head => 0, + }); +} + +sub write_value { + my $self = shift; + my ($obj, $key, $value) = @_; + + my $r = Scalar::Util::reftype( $value ) || ''; + { + last if $r eq ''; + last if $r eq 'HASH'; + last if $r eq 'ARRAY'; + + DBM::Deep->_throw_error( + "Storage of references of type '$r' is not supported." + ); + } + + # This will be a Reference sector + my $sector = $self->_load_sector( $obj->_base_offset ) + or DBM::Deep->_throw_error( "Cannot write to a deleted spot in DBM::Deep." ); + + if ( $sector->staleness != $obj->_staleness ) { + DBM::Deep->_throw_error( "Cannot write to a deleted spot in DBM::Deep." ); + } + + my ($class, $type); + if ( !defined $value ) { + $class = 'DBM::Deep::Engine::Sector::Null'; + } + elsif ( $r eq 'ARRAY' || $r eq 'HASH' ) { + my $tmpvar; + if ( $r eq 'ARRAY' ) { + $tmpvar = tied @$value; + } elsif ( $r eq 'HASH' ) { + $tmpvar = tied %$value; + } + + if ( $tmpvar ) { + my $is_dbm_deep = eval { local $SIG{'__DIE__'}; $tmpvar->isa( 'DBM::Deep' ); }; + + unless ( $is_dbm_deep ) { + DBM::Deep->_throw_error( "Cannot store something that is tied." ); + } + + unless ( $tmpvar->_engine->storage == $self->storage ) { + DBM::Deep->_throw_error( "Cannot store values across DBM::Deep files. Please use export() instead." ); + } + + # First, verify if we're storing the same thing to this spot. If we are, then + # this should be a no-op. -EJS, 2008-05-19 + my $loc = $sector->get_data_location_for({ + key_md5 => $self->_apply_digest( $key ), + allow_head => 1, + }); + + if ( defined($loc) && $loc == $tmpvar->_base_offset ) { + return 1; + } + + #XXX Can this use $loc? + my $value_sector = $self->_load_sector( $tmpvar->_base_offset ); + $sector->write_data({ + key => $key, + key_md5 => $self->_apply_digest( $key ), + value => $value_sector, + }); + $value_sector->increment_refcount; + + return 1; + } + + $class = 'DBM::Deep::Engine::Sector::Reference'; + $type = substr( $r, 0, 1 ); + } + else { + if ( tied($value) ) { + DBM::Deep->_throw_error( "Cannot store something that is tied." ); + } + $class = 'DBM::Deep::Engine::Sector::Scalar'; + } + + # Create this after loading the reference sector in case something bad happens. + # This way, we won't allocate value sector(s) needlessly. + my $value_sector = $class->new({ + engine => $self, + data => $value, + type => $type, + }); + + $sector->write_data({ + key => $key, + key_md5 => $self->_apply_digest( $key ), + value => $value_sector, + }); + + # This code is to make sure we write all the values in the $value to the disk + # and to make sure all changes to $value after the assignment are reflected + # on disk. This may be counter-intuitive at first, but it is correct dwimmery. + # NOTE - simply tying $value won't perform a STORE on each value. Hence, the + # copy to a temp value. + if ( $r eq 'ARRAY' ) { + my @temp = @$value; + tie @$value, 'DBM::Deep', { + base_offset => $value_sector->offset, + staleness => $value_sector->staleness, + storage => $self->storage, + engine => $self, + }; + @$value = @temp; + bless $value, 'DBM::Deep::Array' unless Scalar::Util::blessed( $value ); + } + elsif ( $r eq 'HASH' ) { + my %temp = %$value; + tie %$value, 'DBM::Deep', { + base_offset => $value_sector->offset, + staleness => $value_sector->staleness, + storage => $self->storage, + engine => $self, + }; + + %$value = %temp; + bless $value, 'DBM::Deep::Hash' unless Scalar::Util::blessed( $value ); + } + + return 1; +} + +# XXX Add staleness here +sub get_next_key { + my $self = shift; + my ($obj, $prev_key) = @_; + + # XXX Need to add logic about resetting the iterator if any key in the reference has changed + unless ( $prev_key ) { + $obj->{iterator} = DBM::Deep::Iterator->new({ + base_offset => $obj->_base_offset, + engine => $self, + }); + } + + return $obj->{iterator}->get_next_key( $obj ); +} + +################################################################################ + +sub setup_fh { + my $self = shift; + my ($obj) = @_; + + # We're opening the file. + unless ( $obj->_base_offset ) { + my $bytes_read = $self->_read_file_header; + + # Creating a new file + unless ( $bytes_read ) { + $self->_write_file_header; + + # 1) Create Array/Hash entry + my $initial_reference = DBM::Deep::Engine::Sector::Reference->new({ + engine => $self, + type => $obj->_type, + }); + $obj->{base_offset} = $initial_reference->offset; + $obj->{staleness} = $initial_reference->staleness; + + $self->storage->flush; + } + # Reading from an existing file + else { + $obj->{base_offset} = $bytes_read; + my $initial_reference = DBM::Deep::Engine::Sector::Reference->new({ + engine => $self, + offset => $obj->_base_offset, + }); + unless ( $initial_reference ) { + DBM::Deep->_throw_error("Corrupted file, no master index record"); + } + + unless ($obj->_type eq $initial_reference->type) { + DBM::Deep->_throw_error("File type mismatch"); + } + + $obj->{staleness} = $initial_reference->staleness; + } + } + + return 1; +} + +sub begin_work { + my $self = shift; + my ($obj) = @_; + + if ( $self->trans_id ) { + DBM::Deep->_throw_error( "Cannot begin_work within an active transaction" ); + } + + my @slots = $self->read_txn_slots; + my $found; + for my $i ( 0 .. $#slots ) { + next if $slots[$i]; + + $slots[$i] = 1; + $self->set_trans_id( $i + 1 ); + $found = 1; + last; + } + unless ( $found ) { + DBM::Deep->_throw_error( "Cannot allocate transaction ID" ); + } + $self->write_txn_slots( @slots ); + + if ( !$self->trans_id ) { + DBM::Deep->_throw_error( "Cannot begin_work - no available transactions" ); + } + + return; +} + +sub rollback { + my $self = shift; + my ($obj) = @_; + + if ( !$self->trans_id ) { + DBM::Deep->_throw_error( "Cannot rollback without an active transaction" ); + } + + # Each entry is the file location for a bucket that has a modification for + # this transaction. The entries need to be expunged. + foreach my $entry (@{ $self->get_entries } ) { + # Remove the entry here + my $read_loc = $entry + + $self->hash_size + + $self->byte_size + + $self->byte_size + + ($self->trans_id - 1) * ( $self->byte_size + $STALE_SIZE ); + + my $data_loc = $self->storage->read_at( $read_loc, $self->byte_size ); + $data_loc = unpack( $StP{$self->byte_size}, $data_loc ); + $self->storage->print_at( $read_loc, pack( $StP{$self->byte_size}, 0 ) ); + + if ( $data_loc > 1 ) { + $self->_load_sector( $data_loc )->free; + } + } + + $self->clear_entries; + + my @slots = $self->read_txn_slots; + $slots[$self->trans_id-1] = 0; + $self->write_txn_slots( @slots ); + $self->inc_txn_staleness_counter( $self->trans_id ); + $self->set_trans_id( 0 ); + + return 1; +} + +sub commit { + my $self = shift; + my ($obj) = @_; + + if ( !$self->trans_id ) { + DBM::Deep->_throw_error( "Cannot commit without an active transaction" ); + } + + foreach my $entry (@{ $self->get_entries } ) { + # Overwrite the entry in head with the entry in trans_id + my $base = $entry + + $self->hash_size + + $self->byte_size; + + my $head_loc = $self->storage->read_at( $base, $self->byte_size ); + $head_loc = unpack( $StP{$self->byte_size}, $head_loc ); + + my $spot = $base + $self->byte_size + ($self->trans_id - 1) * ( $self->byte_size + $STALE_SIZE ); + my $trans_loc = $self->storage->read_at( + $spot, $self->byte_size, + ); + + $self->storage->print_at( $base, $trans_loc ); + $self->storage->print_at( + $spot, + pack( $StP{$self->byte_size} . ' ' . $StP{$STALE_SIZE}, (0) x 2 ), + ); + + if ( $head_loc > 1 ) { + $self->_load_sector( $head_loc )->free; + } + } + + $self->clear_entries; + + my @slots = $self->read_txn_slots; + $slots[$self->trans_id-1] = 0; + $self->write_txn_slots( @slots ); + $self->inc_txn_staleness_counter( $self->trans_id ); + $self->set_trans_id( 0 ); + + return 1; +} + +sub read_txn_slots { + my $self = shift; + my $bl = $self->txn_bitfield_len; + my $num_bits = $bl * 8; + return split '', unpack( 'b'.$num_bits, + $self->storage->read_at( + $self->trans_loc, $bl, + ) + ); +} + +sub write_txn_slots { + my $self = shift; + my $num_bits = $self->txn_bitfield_len * 8; + $self->storage->print_at( $self->trans_loc, + pack( 'b'.$num_bits, join('', @_) ), + ); +} + +sub get_running_txn_ids { + my $self = shift; + my @transactions = $self->read_txn_slots; + my @trans_ids = map { $_+1} grep { $transactions[$_] } 0 .. $#transactions; +} + +sub get_txn_staleness_counter { + my $self = shift; + my ($trans_id) = @_; + + # Hardcode staleness of 0 for the HEAD + return 0 unless $trans_id; + + return unpack( $StP{$STALE_SIZE}, + $self->storage->read_at( + $self->trans_loc + $self->txn_bitfield_len + $STALE_SIZE * ($trans_id - 1), + $STALE_SIZE, + ) + ); +} + +sub inc_txn_staleness_counter { + my $self = shift; + my ($trans_id) = @_; + + # Hardcode staleness of 0 for the HEAD + return 0 unless $trans_id; + + $self->storage->print_at( + $self->trans_loc + $self->txn_bitfield_len + $STALE_SIZE * ($trans_id - 1), + pack( $StP{$STALE_SIZE}, $self->get_txn_staleness_counter( $trans_id ) + 1 ), + ); +} + +sub get_entries { + my $self = shift; + return [ keys %{ $self->{entries}{$self->trans_id} ||= {} } ]; +} + +sub add_entry { + my $self = shift; + my ($trans_id, $loc) = @_; + + $self->{entries}{$trans_id} ||= {}; + $self->{entries}{$trans_id}{$loc} = undef; +} + +# If the buckets are being relocated because of a reindexing, the entries +# mechanism needs to be made aware of it. +sub reindex_entry { + my $self = shift; + my ($old_loc, $new_loc) = @_; + + TRANS: + while ( my ($trans_id, $locs) = each %{ $self->{entries} } ) { + if ( exists $locs->{$old_loc} ) { + delete $locs->{$old_loc}; + $locs->{$new_loc} = undef; + next TRANS; + } + } +} + +sub clear_entries { + my $self = shift; + delete $self->{entries}{$self->trans_id}; +} + +################################################################################ + +{ + my $header_fixed = length( SIG_FILE ) + 1 + 4 + 4; + my $this_file_version = 3; + + sub _write_file_header { + my $self = shift; + + my $nt = $self->num_txns; + my $bl = $self->txn_bitfield_len; + + my $header_var = 1 + 1 + 1 + 1 + $bl + $STALE_SIZE * ($nt - 1) + 3 * $self->byte_size; + + my $loc = $self->storage->request_space( $header_fixed + $header_var ); + + $self->storage->print_at( $loc, + SIG_FILE, + SIG_HEADER, + pack('N', $this_file_version), # At this point, we're at 9 bytes + pack('N', $header_var), # header size + # --- Above is $header_fixed. Below is $header_var + pack('C', $self->byte_size), + + # These shenanigans are to allow a 256 within a C + pack('C', $self->max_buckets - 1), + pack('C', $self->data_sector_size - 1), + + pack('C', $nt), + pack('C' . $bl, 0 ), # Transaction activeness bitfield + pack($StP{$STALE_SIZE}.($nt-1), 0 x ($nt-1) ), # Transaction staleness counters + pack($StP{$self->byte_size}, 0), # Start of free chain (blist size) + pack($StP{$self->byte_size}, 0), # Start of free chain (data size) + pack($StP{$self->byte_size}, 0), # Start of free chain (index size) + ); + + #XXX Set these less fragilely + $self->set_trans_loc( $header_fixed + 4 ); + $self->set_chains_loc( $header_fixed + 4 + $bl + $STALE_SIZE * ($nt-1) ); + + return; + } + + sub _read_file_header { + my $self = shift; + + my $buffer = $self->storage->read_at( 0, $header_fixed ); + return unless length($buffer); + + my ($file_signature, $sig_header, $file_version, $size) = unpack( + 'A4 A N N', $buffer + ); + + unless ( $file_signature eq SIG_FILE ) { + $self->storage->close; + DBM::Deep->_throw_error( "Signature not found -- file is not a Deep DB" ); + } + + unless ( $sig_header eq SIG_HEADER ) { + $self->storage->close; + DBM::Deep->_throw_error( "Pre-1.00 file version found" ); + } + + unless ( $file_version == $this_file_version ) { + $self->storage->close; + DBM::Deep->_throw_error( + "Wrong file version found - " . $file_version . + " - expected " . $this_file_version + ); + } + + my $buffer2 = $self->storage->read_at( undef, $size ); + my @values = unpack( 'C C C C', $buffer2 ); + + if ( @values != 4 || grep { !defined } @values ) { + $self->storage->close; + DBM::Deep->_throw_error("Corrupted file - bad header"); + } + + #XXX Add warnings if values weren't set right + @{$self}{qw(byte_size max_buckets data_sector_size num_txns)} = @values; + + # These shenangians are to allow a 256 within a C + $self->{max_buckets} += 1; + $self->{data_sector_size} += 1; + + my $bl = $self->txn_bitfield_len; + + my $header_var = scalar(@values) + $bl + $STALE_SIZE * ($self->num_txns - 1) + 3 * $self->byte_size; + unless ( $size == $header_var ) { + $self->storage->close; + DBM::Deep->_throw_error( "Unexpected size found ($size <-> $header_var)." ); + } + + $self->set_trans_loc( $header_fixed + scalar(@values) ); + $self->set_chains_loc( $header_fixed + scalar(@values) + $bl + $STALE_SIZE * ($self->num_txns - 1) ); + + return length($buffer) + length($buffer2); + } +} + +sub _load_sector { + my $self = shift; + my ($offset) = @_; + + # Add a catch for offset of 0 or 1 + return if !$offset || $offset <= 1; + + my $type = $self->storage->read_at( $offset, 1 ); + return if $type eq chr(0); + + if ( $type eq $self->SIG_ARRAY || $type eq $self->SIG_HASH ) { + return DBM::Deep::Engine::Sector::Reference->new({ + engine => $self, + type => $type, + offset => $offset, + }); + } + # XXX Don't we need key_md5 here? + elsif ( $type eq $self->SIG_BLIST ) { + return DBM::Deep::Engine::Sector::BucketList->new({ + engine => $self, + type => $type, + offset => $offset, + }); + } + elsif ( $type eq $self->SIG_INDEX ) { + return DBM::Deep::Engine::Sector::Index->new({ + engine => $self, + type => $type, + offset => $offset, + }); + } + elsif ( $type eq $self->SIG_NULL ) { + return DBM::Deep::Engine::Sector::Null->new({ + engine => $self, + type => $type, + offset => $offset, + }); + } + elsif ( $type eq $self->SIG_DATA ) { + return DBM::Deep::Engine::Sector::Scalar->new({ + engine => $self, + type => $type, + offset => $offset, + }); + } + # This was deleted from under us, so just return and let the caller figure it out. + elsif ( $type eq $self->SIG_FREE ) { + return; + } + + DBM::Deep->_throw_error( "'$offset': Don't know what to do with type '$type'" ); +} + +sub _apply_digest { + my $self = shift; + return $self->{digest}->(@_); +} + +sub _add_free_blist_sector { shift->_add_free_sector( 0, @_ ) } +sub _add_free_data_sector { shift->_add_free_sector( 1, @_ ) } +sub _add_free_index_sector { shift->_add_free_sector( 2, @_ ) } + +sub _add_free_sector { + my $self = shift; + my ($multiple, $offset, $size) = @_; + + my $chains_offset = $multiple * $self->byte_size; + + my $storage = $self->storage; + + # Increment staleness. + # XXX Can this increment+modulo be done by "&= 0x1" ? + my $staleness = unpack( $StP{$STALE_SIZE}, $storage->read_at( $offset + SIG_SIZE, $STALE_SIZE ) ); + $staleness = ($staleness + 1 ) % ( 2 ** ( 8 * $STALE_SIZE ) ); + $storage->print_at( $offset + SIG_SIZE, pack( $StP{$STALE_SIZE}, $staleness ) ); + + my $old_head = $storage->read_at( $self->chains_loc + $chains_offset, $self->byte_size ); + + $storage->print_at( $self->chains_loc + $chains_offset, + pack( $StP{$self->byte_size}, $offset ), + ); + + # Record the old head in the new sector after the signature and staleness counter + $storage->print_at( $offset + SIG_SIZE + $STALE_SIZE, $old_head ); +} + +sub _request_blist_sector { shift->_request_sector( 0, @_ ) } +sub _request_data_sector { shift->_request_sector( 1, @_ ) } +sub _request_index_sector { shift->_request_sector( 2, @_ ) } + +sub _request_sector { + my $self = shift; + my ($multiple, $size) = @_; + + my $chains_offset = $multiple * $self->byte_size; + + my $old_head = $self->storage->read_at( $self->chains_loc + $chains_offset, $self->byte_size ); + my $loc = unpack( $StP{$self->byte_size}, $old_head ); + + # We don't have any free sectors of the right size, so allocate a new one. + unless ( $loc ) { + my $offset = $self->storage->request_space( $size ); + + # Zero out the new sector. This also guarantees correct increases + # in the filesize. + $self->storage->print_at( $offset, chr(0) x $size ); + + return $offset; + } + + # Read the new head after the signature and the staleness counter + my $new_head = $self->storage->read_at( $loc + SIG_SIZE + $STALE_SIZE, $self->byte_size ); + $self->storage->print_at( $self->chains_loc + $chains_offset, $new_head ); + $self->storage->print_at( + $loc + SIG_SIZE + $STALE_SIZE, + pack( $StP{$self->byte_size}, 0 ), + ); + + return $loc; +} + +################################################################################ + +sub storage { $_[0]{storage} } +sub byte_size { $_[0]{byte_size} } +sub hash_size { $_[0]{hash_size} } +sub hash_chars { $_[0]{hash_chars} } +sub num_txns { $_[0]{num_txns} } +sub max_buckets { $_[0]{max_buckets} } +sub blank_md5 { chr(0) x $_[0]->hash_size } +sub data_sector_size { $_[0]{data_sector_size} } + +# This is a calculated value +sub txn_bitfield_len { + my $self = shift; + unless ( exists $self->{txn_bitfield_len} ) { + my $temp = ($self->num_txns) / 8; + if ( $temp > int( $temp ) ) { + $temp = int( $temp ) + 1; + } + $self->{txn_bitfield_len} = $temp; + } + return $self->{txn_bitfield_len}; +} + +sub trans_id { $_[0]{trans_id} } +sub set_trans_id { $_[0]{trans_id} = $_[1] } + +sub trans_loc { $_[0]{trans_loc} } +sub set_trans_loc { $_[0]{trans_loc} = $_[1] } + +sub chains_loc { $_[0]{chains_loc} } +sub set_chains_loc { $_[0]{chains_loc} = $_[1] } + +sub cache { $_[0]{cache} ||= {} } +sub clear_cache { %{$_[0]->cache} = () } + +sub _dump_file { + my $self = shift; + + # Read the header + my $spot = $self->_read_file_header(); + + my %types = ( + 0 => 'B', + 1 => 'D', + 2 => 'I', + ); + + my %sizes = ( + 'D' => $self->data_sector_size, + 'B' => DBM::Deep::Engine::Sector::BucketList->new({engine=>$self,offset=>1})->size, + 'I' => DBM::Deep::Engine::Sector::Index->new({engine=>$self,offset=>1})->size, + ); + + my $return = ""; + + # Header values + $return .= "NumTxns: " . $self->num_txns . $/; + + # Read the free sector chains + my %sectors; + foreach my $multiple ( 0 .. 2 ) { + $return .= "Chains($types{$multiple}):"; + my $old_loc = $self->chains_loc + $multiple * $self->byte_size; + while ( 1 ) { + my $loc = unpack( + $StP{$self->byte_size}, + $self->storage->read_at( $old_loc, $self->byte_size ), + ); + + # We're now out of free sectors of this kind. + unless ( $loc ) { + last; + } + + $sectors{ $types{$multiple} }{ $loc } = undef; + $old_loc = $loc + SIG_SIZE + $STALE_SIZE; + $return .= " $loc"; + } + $return .= $/; + } + + SECTOR: + while ( $spot < $self->storage->{end} ) { + # Read each sector in order. + my $sector = $self->_load_sector( $spot ); + if ( !$sector ) { + # Find it in the free-sectors that were found already + foreach my $type ( keys %sectors ) { + if ( exists $sectors{$type}{$spot} ) { + my $size = $sizes{$type}; + $return .= sprintf "%08d: %s %04d\n", $spot, 'F' . $type, $size; + $spot += $size; + next SECTOR; + } + } + + die "********\n$return\nDidn't find free sector for $spot in chains\n********\n"; + } + else { + $return .= sprintf "%08d: %s %04d", $spot, $sector->type, $sector->size; + if ( $sector->type eq 'D' ) { + $return .= ' ' . $sector->data; + } + elsif ( $sector->type eq 'A' || $sector->type eq 'H' ) { + $return .= ' REF: ' . $sector->get_refcount; + } + elsif ( $sector->type eq 'B' ) { + foreach my $bucket ( $sector->chopped_up ) { + $return .= "\n "; + $return .= sprintf "%08d", unpack($StP{$self->byte_size}, + substr( $bucket->[-1], $self->hash_size, $self->byte_size), + ); + my $l = unpack( $StP{$self->byte_size}, + substr( $bucket->[-1], + $self->hash_size + $self->byte_size, + $self->byte_size, + ), + ); + $return .= sprintf " %08d", $l; + foreach my $txn ( 0 .. $self->num_txns - 2 ) { + my $l = unpack( $StP{$self->byte_size}, + substr( $bucket->[-1], + $self->hash_size + 2 * $self->byte_size + $txn * ($self->byte_size + $STALE_SIZE), + $self->byte_size, + ), + ); + $return .= sprintf " %08d", $l; + } + } + } + $return .= $/; + + $spot += $sector->size; + } + } + + return $return; +} + +################################################################################ + +package DBM::Deep::Iterator; + +sub new { + my $class = shift; + my ($args) = @_; + + my $self = bless { + breadcrumbs => [], + engine => $args->{engine}, + base_offset => $args->{base_offset}, + }, $class; + + Scalar::Util::weaken( $self->{engine} ); + + return $self; +} + +sub reset { $_[0]{breadcrumbs} = [] } + +sub get_sector_iterator { + my $self = shift; + my ($loc) = @_; + + my $sector = $self->{engine}->_load_sector( $loc ) + or return; + + if ( $sector->isa( 'DBM::Deep::Engine::Sector::Index' ) ) { + return DBM::Deep::Iterator::Index->new({ + iterator => $self, + sector => $sector, + }); + } + elsif ( $sector->isa( 'DBM::Deep::Engine::Sector::BucketList' ) ) { + return DBM::Deep::Iterator::BucketList->new({ + iterator => $self, + sector => $sector, + }); + } + + DBM::Deep->_throw_error( "get_sector_iterator(): Why did $loc make a $sector?" ); +} + +sub get_next_key { + my $self = shift; + my ($obj) = @_; + + my $crumbs = $self->{breadcrumbs}; + my $e = $self->{engine}; + + unless ( @$crumbs ) { + # This will be a Reference sector + my $sector = $e->_load_sector( $self->{base_offset} ) + # If no sector is found, thist must have been deleted from under us. + or return; + + if ( $sector->staleness != $obj->_staleness ) { + return; + } + + my $loc = $sector->get_blist_loc + or return; + + push @$crumbs, $self->get_sector_iterator( $loc ); + } + + FIND_NEXT_KEY: { + # We're at the end. + unless ( @$crumbs ) { + $self->reset; + return; + } + + my $iterator = $crumbs->[-1]; + + # This level is done. + if ( $iterator->at_end ) { + pop @$crumbs; + redo FIND_NEXT_KEY; + } + + if ( $iterator->isa( 'DBM::Deep::Iterator::Index' ) ) { + # If we don't have any more, it will be caught at the + # prior check. + if ( my $next = $iterator->get_next_iterator ) { + push @$crumbs, $next; + } + redo FIND_NEXT_KEY; + } + + unless ( $iterator->isa( 'DBM::Deep::Iterator::BucketList' ) ) { + DBM::Deep->_throw_error( + "Should have a bucketlist iterator here - instead have $iterator" + ); + } + + # At this point, we have a BucketList iterator + my $key = $iterator->get_next_key; + if ( defined $key ) { + return $key; + } + #XXX else { $iterator->set_to_end() } ? + + # We hit the end of the bucketlist iterator, so redo + redo FIND_NEXT_KEY; + } + + DBM::Deep->_throw_error( "get_next_key(): How did we get here?" ); +} + +package DBM::Deep::Iterator::Index; + +sub new { + my $self = bless $_[1] => $_[0]; + $self->{curr_index} = 0; + return $self; +} + +sub at_end { + my $self = shift; + return $self->{curr_index} >= $self->{iterator}{engine}->hash_chars; +} + +sub get_next_iterator { + my $self = shift; + + my $loc; + while ( !$loc ) { + return if $self->at_end; + $loc = $self->{sector}->get_entry( $self->{curr_index}++ ); + } + + return $self->{iterator}->get_sector_iterator( $loc ); +} + +package DBM::Deep::Iterator::BucketList; + +sub new { + my $self = bless $_[1] => $_[0]; + $self->{curr_index} = 0; + return $self; +} + +sub at_end { + my $self = shift; + return $self->{curr_index} >= $self->{iterator}{engine}->max_buckets; +} + +sub get_next_key { + my $self = shift; + + return if $self->at_end; + + my $idx = $self->{curr_index}++; + + my $data_loc = $self->{sector}->get_data_location_for({ + allow_head => 1, + idx => $idx, + }) or return; + + #XXX Do we want to add corruption checks here? + return $self->{sector}->get_key_for( $idx )->data; +} + +package DBM::Deep::Engine::Sector; + +sub new { + my $self = bless $_[1], $_[0]; + Scalar::Util::weaken( $self->{engine} ); + $self->_init; + return $self; +} + +#sub _init {} +#sub clone { DBM::Deep->_throw_error( "Must be implemented in the child class" ); } + +sub engine { $_[0]{engine} } +sub offset { $_[0]{offset} } +sub type { $_[0]{type} } + +sub base_size { + my $self = shift; + return $self->engine->SIG_SIZE + $STALE_SIZE; +} + +sub free { + my $self = shift; + + my $e = $self->engine; + + $e->storage->print_at( $self->offset, $e->SIG_FREE ); + # Skip staleness counter + $e->storage->print_at( $self->offset + $self->base_size, + chr(0) x ($self->size - $self->base_size), + ); + + my $free_meth = $self->free_meth; + $e->$free_meth( $self->offset, $self->size ); + + return; +} + +package DBM::Deep::Engine::Sector::Data; + +our @ISA = qw( DBM::Deep::Engine::Sector ); + +# This is in bytes +sub size { $_[0]{engine}->data_sector_size } +sub free_meth { return '_add_free_data_sector' } + +sub clone { + my $self = shift; + return ref($self)->new({ + engine => $self->engine, + type => $self->type, + data => $self->data, + }); +} + +package DBM::Deep::Engine::Sector::Scalar; + +our @ISA = qw( DBM::Deep::Engine::Sector::Data ); + +sub free { + my $self = shift; + + my $chain_loc = $self->chain_loc; + + $self->SUPER::free(); + + if ( $chain_loc ) { + $self->engine->_load_sector( $chain_loc )->free; + } + + return; +} + +sub type { $_[0]{engine}->SIG_DATA } +sub _init { + my $self = shift; + + my $engine = $self->engine; + + unless ( $self->offset ) { + my $data_section = $self->size - $self->base_size - $engine->byte_size - 1; + + $self->{offset} = $engine->_request_data_sector( $self->size ); + + my $data = delete $self->{data}; + my $dlen = length $data; + my $continue = 1; + my $curr_offset = $self->offset; + while ( $continue ) { + + my $next_offset = 0; + + my ($leftover, $this_len, $chunk); + if ( $dlen > $data_section ) { + $leftover = 0; + $this_len = $data_section; + $chunk = substr( $data, 0, $this_len ); + + $dlen -= $data_section; + $next_offset = $engine->_request_data_sector( $self->size ); + $data = substr( $data, $this_len ); + } + else { + $leftover = $data_section - $dlen; + $this_len = $dlen; + $chunk = $data; + + $continue = 0; + } + + $engine->storage->print_at( $curr_offset, $self->type ); # Sector type + # Skip staleness + $engine->storage->print_at( $curr_offset + $self->base_size, + pack( $StP{$engine->byte_size}, $next_offset ), # Chain loc + pack( $StP{1}, $this_len ), # Data length + $chunk, # Data to be stored in this sector + chr(0) x $leftover, # Zero-fill the rest + ); + + $curr_offset = $next_offset; + } + + return; + } +} + +sub data_length { + my $self = shift; + + my $buffer = $self->engine->storage->read_at( + $self->offset + $self->base_size + $self->engine->byte_size, 1 + ); + + return unpack( $StP{1}, $buffer ); +} + +sub chain_loc { + my $self = shift; + return unpack( + $StP{$self->engine->byte_size}, + $self->engine->storage->read_at( + $self->offset + $self->base_size, + $self->engine->byte_size, + ), + ); +} + +sub data { + my $self = shift; +# my ($args) = @_; +# $args ||= {}; + + my $data; + while ( 1 ) { + my $chain_loc = $self->chain_loc; + + $data .= $self->engine->storage->read_at( + $self->offset + $self->base_size + $self->engine->byte_size + 1, $self->data_length, + ); + + last unless $chain_loc; + + $self = $self->engine->_load_sector( $chain_loc ); + } + + return $data; +} + +package DBM::Deep::Engine::Sector::Null; + +our @ISA = qw( DBM::Deep::Engine::Sector::Data ); + +sub type { $_[0]{engine}->SIG_NULL } +sub data_length { 0 } +sub data { return } + +sub _init { + my $self = shift; + + my $engine = $self->engine; + + unless ( $self->offset ) { + my $leftover = $self->size - $self->base_size - 1 * $engine->byte_size - 1; + + $self->{offset} = $engine->_request_data_sector( $self->size ); + $engine->storage->print_at( $self->offset, $self->type ); # Sector type + # Skip staleness counter + $engine->storage->print_at( $self->offset + $self->base_size, + pack( $StP{$engine->byte_size}, 0 ), # Chain loc + pack( $StP{1}, $self->data_length ), # Data length + chr(0) x $leftover, # Zero-fill the rest + ); + + return; + } +} + +package DBM::Deep::Engine::Sector::Reference; + +our @ISA = qw( DBM::Deep::Engine::Sector::Data ); + +sub _init { + my $self = shift; + + my $e = $self->engine; + + unless ( $self->offset ) { + my $classname = Scalar::Util::blessed( delete $self->{data} ); + my $leftover = $self->size - $self->base_size - 3 * $e->byte_size; + + my $class_offset = 0; + if ( defined $classname ) { + my $class_sector = DBM::Deep::Engine::Sector::Scalar->new({ + engine => $e, + data => $classname, + }); + $class_offset = $class_sector->offset; + } + + $self->{offset} = $e->_request_data_sector( $self->size ); + $e->storage->print_at( $self->offset, $self->type ); # Sector type + # Skip staleness counter + $e->storage->print_at( $self->offset + $self->base_size, + pack( $StP{$e->byte_size}, 0 ), # Index/BList loc + pack( $StP{$e->byte_size}, $class_offset ), # Classname loc + pack( $StP{$e->byte_size}, 1 ), # Initial refcount + chr(0) x $leftover, # Zero-fill the rest + ); + } + else { + $self->{type} = $e->storage->read_at( $self->offset, 1 ); + } + + $self->{staleness} = unpack( + $StP{$STALE_SIZE}, + $e->storage->read_at( $self->offset + $e->SIG_SIZE, $STALE_SIZE ), + ); + + return; +} + +sub staleness { $_[0]{staleness} } + +sub get_data_location_for { + my $self = shift; + my ($args) = @_; + + # Assume that the head is not allowed unless otherwise specified. + $args->{allow_head} = 0 unless exists $args->{allow_head}; + + # Assume we don't create a new blist location unless otherwise specified. + $args->{create} = 0 unless exists $args->{create}; + + my $blist = $self->get_bucket_list({ + key_md5 => $args->{key_md5}, + key => $args->{key}, + create => $args->{create}, + }); + return unless $blist && $blist->{found}; + + # At this point, $blist knows where the md5 is. What it -doesn't- know yet + # is whether or not this transaction has this key. That's part of the next + # function call. + my $location = $blist->get_data_location_for({ + allow_head => $args->{allow_head}, + }) or return; + + return $location; +} + +sub get_data_for { + my $self = shift; + my ($args) = @_; + + my $location = $self->get_data_location_for( $args ) + or return; + + return $self->engine->_load_sector( $location ); +} + +sub write_data { + my $self = shift; + my ($args) = @_; + + my $blist = $self->get_bucket_list({ + key_md5 => $args->{key_md5}, + key => $args->{key}, + create => 1, + }) or DBM::Deep->_throw_error( "How did write_data fail (no blist)?!" ); + + # Handle any transactional bookkeeping. + if ( $self->engine->trans_id ) { + if ( ! $blist->has_md5 ) { + $blist->mark_deleted({ + trans_id => 0, + }); + } + } + else { + my @trans_ids = $self->engine->get_running_txn_ids; + if ( $blist->has_md5 ) { + if ( @trans_ids ) { + my $old_value = $blist->get_data_for; + foreach my $other_trans_id ( @trans_ids ) { + next if $blist->get_data_location_for({ + trans_id => $other_trans_id, + allow_head => 0, + }); + $blist->write_md5({ + trans_id => $other_trans_id, + key => $args->{key}, + key_md5 => $args->{key_md5}, + value => $old_value->clone, + }); + } + } + } + else { + if ( @trans_ids ) { + foreach my $other_trans_id ( @trans_ids ) { + #XXX This doesn't seem to possible to ever happen . . . + next if $blist->get_data_location_for({ trans_id => $other_trans_id, allow_head => 0 }); + $blist->mark_deleted({ + trans_id => $other_trans_id, + }); + } + } + } + } + + #XXX Is this safe to do transactionally? + # Free the place we're about to write to. + if ( $blist->get_data_location_for({ allow_head => 0 }) ) { + $blist->get_data_for({ allow_head => 0 })->free; + } + + $blist->write_md5({ + key => $args->{key}, + key_md5 => $args->{key_md5}, + value => $args->{value}, + }); +} + +sub delete_key { + my $self = shift; + my ($args) = @_; + + # XXX What should happen if this fails? + my $blist = $self->get_bucket_list({ + key_md5 => $args->{key_md5}, + }) or DBM::Deep->_throw_error( "How did delete_key fail (no blist)?!" ); + + # Save the location so that we can free the data + my $location = $blist->get_data_location_for({ + allow_head => 0, + }); + my $old_value = $location && $self->engine->_load_sector( $location ); + + my @trans_ids = $self->engine->get_running_txn_ids; + + # If we're the HEAD and there are running txns, then we need to clone this value to the other + # transactions to preserve Isolation. + if ( $self->engine->trans_id == 0 ) { + if ( @trans_ids ) { + foreach my $other_trans_id ( @trans_ids ) { + next if $blist->get_data_location_for({ trans_id => $other_trans_id, allow_head => 0 }); + $blist->write_md5({ + trans_id => $other_trans_id, + key => $args->{key}, + key_md5 => $args->{key_md5}, + value => $old_value->clone, + }); + } + } + } + + my $data; + if ( @trans_ids ) { + $blist->mark_deleted( $args ); + + if ( $old_value ) { + $data = $old_value->data({ export => 1 }); + $old_value->free; + } + } + else { + $data = $blist->delete_md5( $args ); + } + + return $data; +} + +sub get_blist_loc { + my $self = shift; + + my $e = $self->engine; + my $blist_loc = $e->storage->read_at( $self->offset + $self->base_size, $e->byte_size ); + return unpack( $StP{$e->byte_size}, $blist_loc ); +} + +sub get_bucket_list { + my $self = shift; + my ($args) = @_; + $args ||= {}; + + # XXX Add in check here for recycling? + + my $engine = $self->engine; + + my $blist_loc = $self->get_blist_loc; + + # There's no index or blist yet + unless ( $blist_loc ) { + return unless $args->{create}; + + my $blist = DBM::Deep::Engine::Sector::BucketList->new({ + engine => $engine, + key_md5 => $args->{key_md5}, + }); + + $engine->storage->print_at( $self->offset + $self->base_size, + pack( $StP{$engine->byte_size}, $blist->offset ), + ); + + return $blist; + } + + my $sector = $engine->_load_sector( $blist_loc ) + or DBM::Deep->_throw_error( "Cannot read sector at $blist_loc in get_bucket_list()" ); + my $i = 0; + my $last_sector = undef; + while ( $sector->isa( 'DBM::Deep::Engine::Sector::Index' ) ) { + $blist_loc = $sector->get_entry( ord( substr( $args->{key_md5}, $i++, 1 ) ) ); + $last_sector = $sector; + if ( $blist_loc ) { + $sector = $engine->_load_sector( $blist_loc ) + or DBM::Deep->_throw_error( "Cannot read sector at $blist_loc in get_bucket_list()" ); + } + else { + $sector = undef; + last; + } + } + + # This means we went through the Index sector(s) and found an empty slot + unless ( $sector ) { + return unless $args->{create}; + + DBM::Deep->_throw_error( "No last_sector when attempting to build a new entry" ) + unless $last_sector; + + my $blist = DBM::Deep::Engine::Sector::BucketList->new({ + engine => $engine, + key_md5 => $args->{key_md5}, + }); + + $last_sector->set_entry( ord( substr( $args->{key_md5}, $i - 1, 1 ) ) => $blist->offset ); + + return $blist; + } + + $sector->find_md5( $args->{key_md5} ); + + # See whether or not we need to reindex the bucketlist + # Yes, the double-braces are there for a reason. if() doesn't create a redo-able block, + # so we have to create a bare block within the if() for redo-purposes. Patch and idea + # submitted by sprout@cpan.org. -RobK, 2008-01-09 + if ( !$sector->has_md5 && $args->{create} && $sector->{idx} == -1 ) {{ + my $redo; + + my $new_index = DBM::Deep::Engine::Sector::Index->new({ + engine => $engine, + }); + + my %blist_cache; + #XXX q.v. the comments for this function. + foreach my $entry ( $sector->chopped_up ) { + my ($spot, $md5) = @{$entry}; + my $idx = ord( substr( $md5, $i, 1 ) ); + + # XXX This is inefficient + my $blist = $blist_cache{$idx} + ||= DBM::Deep::Engine::Sector::BucketList->new({ + engine => $engine, + }); + + $new_index->set_entry( $idx => $blist->offset ); + + my $new_spot = $blist->write_at_next_open( $md5 ); + $engine->reindex_entry( $spot => $new_spot ); + } + + # Handle the new item separately. + { + my $idx = ord( substr( $args->{key_md5}, $i, 1 ) ); + + # If all the previous blist's items have been thrown into one + # blist and the new item belongs in there too, we need + # another index. + if ( keys %blist_cache == 1 and each %blist_cache == $idx ) { + ++$i, ++$redo; + } else { + my $blist = $blist_cache{$idx} + ||= DBM::Deep::Engine::Sector::BucketList->new({ + engine => $engine, + }); + + $new_index->set_entry( $idx => $blist->offset ); + + #XXX THIS IS HACKY! + $blist->find_md5( $args->{key_md5} ); + $blist->write_md5({ + key => $args->{key}, + key_md5 => $args->{key_md5}, + value => DBM::Deep::Engine::Sector::Null->new({ + engine => $engine, + data => undef, + }), + }); + } +# my $blist = $blist_cache{$idx} +# ||= DBM::Deep::Engine::Sector::BucketList->new({ +# engine => $engine, +# }); +# +# $new_index->set_entry( $idx => $blist->offset ); +# +# #XXX THIS IS HACKY! +# $blist->find_md5( $args->{key_md5} ); +# $blist->write_md5({ +# key => $args->{key}, +# key_md5 => $args->{key_md5}, +# value => DBM::Deep::Engine::Sector::Null->new({ +# engine => $engine, +# data => undef, +# }), +# }); + } + + if ( $last_sector ) { + $last_sector->set_entry( + ord( substr( $args->{key_md5}, $i - 1, 1 ) ), + $new_index->offset, + ); + } else { + $engine->storage->print_at( $self->offset + $self->base_size, + pack( $StP{$engine->byte_size}, $new_index->offset ), + ); + } + + $sector->clear; + $sector->free; + + if ( $redo ) { + (undef, $sector) = %blist_cache; + $last_sector = $new_index; + redo; + } + + $sector = $blist_cache{ ord( substr( $args->{key_md5}, $i, 1 ) ) }; + $sector->find_md5( $args->{key_md5} ); + }} + + return $sector; +} + +sub get_class_offset { + my $self = shift; + + my $e = $self->engine; + return unpack( + $StP{$e->byte_size}, + $e->storage->read_at( + $self->offset + $self->base_size + 1 * $e->byte_size, $e->byte_size, + ), + ); +} + +sub get_classname { + my $self = shift; + + my $class_offset = $self->get_class_offset; + + return unless $class_offset; + + return $self->engine->_load_sector( $class_offset )->data; +} + +sub data { + my $self = shift; + my ($args) = @_; + $args ||= {}; + + my $obj; + unless ( $obj = $self->engine->cache->{ $self->offset } ) { + $obj = DBM::Deep->new({ + type => $self->type, + base_offset => $self->offset, + staleness => $self->staleness, + storage => $self->engine->storage, + engine => $self->engine, + }); + + if ( $self->engine->storage->{autobless} ) { + my $classname = $self->get_classname; + if ( defined $classname ) { + bless $obj, $classname; + } + } + + $self->engine->cache->{$self->offset} = $obj; + } + + # We're not exporting, so just return. + unless ( $args->{export} ) { + return $obj; + } + + # We shouldn't export if this is still referred to. + if ( $self->get_refcount > 1 ) { + return $obj; + } + + return $obj->export; +} + +sub free { + my $self = shift; + + # We're not ready to be removed yet. + if ( $self->decrement_refcount > 0 ) { + return; + } + + # Rebless the object into DBM::Deep::Null. + eval { %{ $self->engine->cache->{ $self->offset } } = (); }; + eval { @{ $self->engine->cache->{ $self->offset } } = (); }; + bless $self->engine->cache->{ $self->offset }, 'DBM::Deep::Null'; + delete $self->engine->cache->{ $self->offset }; + + my $blist_loc = $self->get_blist_loc; + $self->engine->_load_sector( $blist_loc )->free if $blist_loc; + + my $class_loc = $self->get_class_offset; + $self->engine->_load_sector( $class_loc )->free if $class_loc; + + $self->SUPER::free(); +} + +sub increment_refcount { + my $self = shift; + + my $refcount = $self->get_refcount; + + $refcount++; + + $self->write_refcount( $refcount ); + + return $refcount; +} + +sub decrement_refcount { + my $self = shift; + + my $refcount = $self->get_refcount; + + $refcount--; + + $self->write_refcount( $refcount ); + + return $refcount; +} + +sub get_refcount { + my $self = shift; + + my $e = $self->engine; + return unpack( + $StP{$e->byte_size}, + $e->storage->read_at( + $self->offset + $self->base_size + 2 * $e->byte_size, $e->byte_size, + ), + ); +} + +sub write_refcount { + my $self = shift; + my ($num) = @_; + + my $e = $self->engine; + $e->storage->print_at( + $self->offset + $self->base_size + 2 * $e->byte_size, + pack( $StP{$e->byte_size}, $num ), + ); +} + +package DBM::Deep::Engine::Sector::BucketList; + +our @ISA = qw( DBM::Deep::Engine::Sector ); + +sub _init { + my $self = shift; + + my $engine = $self->engine; + + unless ( $self->offset ) { + my $leftover = $self->size - $self->base_size; + + $self->{offset} = $engine->_request_blist_sector( $self->size ); + $engine->storage->print_at( $self->offset, $engine->SIG_BLIST ); # Sector type + # Skip staleness counter + $engine->storage->print_at( $self->offset + $self->base_size, + chr(0) x $leftover, # Zero-fill the data + ); + } + + if ( $self->{key_md5} ) { + $self->find_md5; + } + + return $self; +} + +sub clear { + my $self = shift; + $self->engine->storage->print_at( $self->offset + $self->base_size, + chr(0) x ($self->size - $self->base_size), # Zero-fill the data + ); +} + +sub size { + my $self = shift; + unless ( $self->{size} ) { + my $e = $self->engine; + # Base + numbuckets * bucketsize + $self->{size} = $self->base_size + $e->max_buckets * $self->bucket_size; + } + return $self->{size}; +} + +sub free_meth { return '_add_free_blist_sector' } + +sub free { + my $self = shift; + + my $e = $self->engine; + foreach my $bucket ( $self->chopped_up ) { + my $rest = $bucket->[-1]; + + # Delete the keysector + my $l = unpack( $StP{$e->byte_size}, substr( $rest, $e->hash_size, $e->byte_size ) ); + my $s = $e->_load_sector( $l ); $s->free if $s; + + # Delete the HEAD sector + $l = unpack( $StP{$e->byte_size}, + substr( $rest, + $e->hash_size + $e->byte_size, + $e->byte_size, + ), + ); + $s = $e->_load_sector( $l ); $s->free if $s; + + foreach my $txn ( 0 .. $e->num_txns - 2 ) { + my $l = unpack( $StP{$e->byte_size}, + substr( $rest, + $e->hash_size + 2 * $e->byte_size + $txn * ($e->byte_size + $STALE_SIZE), + $e->byte_size, + ), + ); + my $s = $e->_load_sector( $l ); $s->free if $s; + } + } + + $self->SUPER::free(); +} + +sub bucket_size { + my $self = shift; + unless ( $self->{bucket_size} ) { + my $e = $self->engine; + # Key + head (location) + transactions (location + staleness-counter) + my $location_size = $e->byte_size + $e->byte_size + ($e->num_txns - 1) * ($e->byte_size + $STALE_SIZE); + $self->{bucket_size} = $e->hash_size + $location_size; + } + return $self->{bucket_size}; +} + +# XXX This is such a poor hack. I need to rethink this code. +sub chopped_up { + my $self = shift; + + my $e = $self->engine; + + my @buckets; + foreach my $idx ( 0 .. $e->max_buckets - 1 ) { + my $spot = $self->offset + $self->base_size + $idx * $self->bucket_size; + my $md5 = $e->storage->read_at( $spot, $e->hash_size ); + + #XXX If we're chopping, why would we ever have the blank_md5? + last if $md5 eq $e->blank_md5; + + my $rest = $e->storage->read_at( undef, $self->bucket_size - $e->hash_size ); + push @buckets, [ $spot, $md5 . $rest ]; + } + + return @buckets; +} + +sub write_at_next_open { + my $self = shift; + my ($entry) = @_; + + #XXX This is such a hack! + $self->{_next_open} = 0 unless exists $self->{_next_open}; + + my $spot = $self->offset + $self->base_size + $self->{_next_open}++ * $self->bucket_size; + $self->engine->storage->print_at( $spot, $entry ); + + return $spot; +} + +sub has_md5 { + my $self = shift; + unless ( exists $self->{found} ) { + $self->find_md5; + } + return $self->{found}; +} + +sub find_md5 { + my $self = shift; + + $self->{found} = undef; + $self->{idx} = -1; + + if ( @_ ) { + $self->{key_md5} = shift; + } + + # If we don't have an MD5, then what are we supposed to do? + unless ( exists $self->{key_md5} ) { + DBM::Deep->_throw_error( "Cannot find_md5 without a key_md5 set" ); + } + + my $e = $self->engine; + foreach my $idx ( 0 .. $e->max_buckets - 1 ) { + my $potential = $e->storage->read_at( + $self->offset + $self->base_size + $idx * $self->bucket_size, $e->hash_size, + ); + + if ( $potential eq $e->blank_md5 ) { + $self->{idx} = $idx; + return; + } + + if ( $potential eq $self->{key_md5} ) { + $self->{found} = 1; + $self->{idx} = $idx; + return; + } + } + + return; +} + +sub write_md5 { + my $self = shift; + my ($args) = @_; + + DBM::Deep->_throw_error( "write_md5: no key" ) unless exists $args->{key}; + DBM::Deep->_throw_error( "write_md5: no key_md5" ) unless exists $args->{key_md5}; + DBM::Deep->_throw_error( "write_md5: no value" ) unless exists $args->{value}; + + my $engine = $self->engine; + + $args->{trans_id} = $engine->trans_id unless exists $args->{trans_id}; + + my $spot = $self->offset + $self->base_size + $self->{idx} * $self->bucket_size; + $engine->add_entry( $args->{trans_id}, $spot ); + + unless ($self->{found}) { + my $key_sector = DBM::Deep::Engine::Sector::Scalar->new({ + engine => $engine, + data => $args->{key}, + }); + + $engine->storage->print_at( $spot, + $args->{key_md5}, + pack( $StP{$engine->byte_size}, $key_sector->offset ), + ); + } + + my $loc = $spot + + $engine->hash_size + + $engine->byte_size; + + if ( $args->{trans_id} ) { + $loc += $engine->byte_size + ($args->{trans_id} - 1) * ( $engine->byte_size + $STALE_SIZE ); + + $engine->storage->print_at( $loc, + pack( $StP{$engine->byte_size}, $args->{value}->offset ), + pack( $StP{$STALE_SIZE}, $engine->get_txn_staleness_counter( $args->{trans_id} ) ), + ); + } + else { + $engine->storage->print_at( $loc, + pack( $StP{$engine->byte_size}, $args->{value}->offset ), + ); + } +} + +sub mark_deleted { + my $self = shift; + my ($args) = @_; + $args ||= {}; + + my $engine = $self->engine; + + $args->{trans_id} = $engine->trans_id unless exists $args->{trans_id}; + + my $spot = $self->offset + $self->base_size + $self->{idx} * $self->bucket_size; + $engine->add_entry( $args->{trans_id}, $spot ); + + my $loc = $spot + + $engine->hash_size + + $engine->byte_size; + + if ( $args->{trans_id} ) { + $loc += $engine->byte_size + ($args->{trans_id} - 1) * ( $engine->byte_size + $STALE_SIZE ); + + $engine->storage->print_at( $loc, + pack( $StP{$engine->byte_size}, 1 ), # 1 is the marker for deleted + pack( $StP{$STALE_SIZE}, $engine->get_txn_staleness_counter( $args->{trans_id} ) ), + ); + } + else { + $engine->storage->print_at( $loc, + pack( $StP{$engine->byte_size}, 1 ), # 1 is the marker for deleted + ); + } + +} + +sub delete_md5 { + my $self = shift; + my ($args) = @_; + + my $engine = $self->engine; + return undef unless $self->{found}; + + # Save the location so that we can free the data + my $location = $self->get_data_location_for({ + allow_head => 0, + }); + my $key_sector = $self->get_key_for; + + my $spot = $self->offset + $self->base_size + $self->{idx} * $self->bucket_size; + $engine->storage->print_at( $spot, + $engine->storage->read_at( + $spot + $self->bucket_size, + $self->bucket_size * ( $engine->max_buckets - $self->{idx} - 1 ), + ), + chr(0) x $self->bucket_size, + ); + + $key_sector->free; + + my $data_sector = $self->engine->_load_sector( $location ); + my $data = $data_sector->data({ export => 1 }); + $data_sector->free; + + return $data; +} + +sub get_data_location_for { + my $self = shift; + my ($args) = @_; + $args ||= {}; + + $args->{allow_head} = 0 unless exists $args->{allow_head}; + $args->{trans_id} = $self->engine->trans_id unless exists $args->{trans_id}; + $args->{idx} = $self->{idx} unless exists $args->{idx}; + + my $e = $self->engine; + + my $spot = $self->offset + $self->base_size + + $args->{idx} * $self->bucket_size + + $e->hash_size + + $e->byte_size; + + if ( $args->{trans_id} ) { + $spot += $e->byte_size + ($args->{trans_id} - 1) * ( $e->byte_size + $STALE_SIZE ); + } + + my $buffer = $e->storage->read_at( + $spot, + $e->byte_size + $STALE_SIZE, + ); + my ($loc, $staleness) = unpack( $StP{$e->byte_size} . ' ' . $StP{$STALE_SIZE}, $buffer ); + + # XXX Merge the two if-clauses below + if ( $args->{trans_id} ) { + # We have found an entry that is old, so get rid of it + if ( $staleness != (my $s = $e->get_txn_staleness_counter( $args->{trans_id} ) ) ) { + $e->storage->print_at( + $spot, + pack( $StP{$e->byte_size} . ' ' . $StP{$STALE_SIZE}, (0) x 2 ), + ); + $loc = 0; + } + } + + # If we're in a transaction and we never wrote to this location, try the + # HEAD instead. + if ( $args->{trans_id} && !$loc && $args->{allow_head} ) { + return $self->get_data_location_for({ + trans_id => 0, + allow_head => 1, + idx => $args->{idx}, + }); + } + + return $loc <= 1 ? 0 : $loc; +} + +sub get_data_for { + my $self = shift; + my ($args) = @_; + $args ||= {}; + + return unless $self->{found}; + my $location = $self->get_data_location_for({ + allow_head => $args->{allow_head}, + }); + return $self->engine->_load_sector( $location ); +} + +sub get_key_for { + my $self = shift; + my ($idx) = @_; + $idx = $self->{idx} unless defined $idx; + + if ( $idx >= $self->engine->max_buckets ) { + DBM::Deep->_throw_error( "get_key_for(): Attempting to retrieve $idx" ); + } + + my $location = $self->engine->storage->read_at( + $self->offset + $self->base_size + $idx * $self->bucket_size + $self->engine->hash_size, + $self->engine->byte_size, + ); + $location = unpack( $StP{$self->engine->byte_size}, $location ); + DBM::Deep->_throw_error( "get_key_for: No location?" ) unless $location; + + return $self->engine->_load_sector( $location ); +} + +package DBM::Deep::Engine::Sector::Index; + +our @ISA = qw( DBM::Deep::Engine::Sector ); + +sub _init { + my $self = shift; + + my $engine = $self->engine; + + unless ( $self->offset ) { + my $leftover = $self->size - $self->base_size; + + $self->{offset} = $engine->_request_index_sector( $self->size ); + $engine->storage->print_at( $self->offset, $engine->SIG_INDEX ); # Sector type + # Skip staleness counter + $engine->storage->print_at( $self->offset + $self->base_size, + chr(0) x $leftover, # Zero-fill the rest + ); + } + + return $self; +} + +#XXX Change here +sub size { + my $self = shift; + unless ( $self->{size} ) { + my $e = $self->engine; + $self->{size} = $self->base_size + $e->byte_size * $e->hash_chars; + } + return $self->{size}; +} + +sub free_meth { return '_add_free_index_sector' } + +sub free { + my $self = shift; + my $e = $self->engine; + + for my $i ( 0 .. $e->hash_chars - 1 ) { + my $l = $self->get_entry( $i ) or next; + $e->_load_sector( $l )->free; + } + + $self->SUPER::free(); +} + +sub _loc_for { + my $self = shift; + my ($idx) = @_; + return $self->offset + $self->base_size + $idx * $self->engine->byte_size; +} + +sub get_entry { + my $self = shift; + my ($idx) = @_; + + my $e = $self->engine; + + DBM::Deep->_throw_error( "get_entry: Out of range ($idx)" ) + if $idx < 0 || $idx >= $e->hash_chars; + + return unpack( + $StP{$e->byte_size}, + $e->storage->read_at( $self->_loc_for( $idx ), $e->byte_size ), + ); +} + +sub set_entry { + my $self = shift; + my ($idx, $loc) = @_; + + my $e = $self->engine; + + DBM::Deep->_throw_error( "set_entry: Out of range ($idx)" ) + if $idx < 0 || $idx >= $e->hash_chars; + + $self->engine->storage->print_at( + $self->_loc_for( $idx ), + pack( $StP{$e->byte_size}, $loc ), + ); +} + +# This was copied from MARCEL's Class::Null. However, I couldn't use it because +# I need an undef value, not an implementation of the Null Class pattern. +package DBM::Deep::Null; + +use overload + 'bool' => sub { undef }, + '""' => sub { undef }, + '0+' => sub { undef }, + fallback => 1, + nomethod => 'AUTOLOAD'; + +sub AUTOLOAD { return; } + +1; +__END__ diff --git a/Master/tlpkg/tlperl/lib/DBM/Deep/File.pm b/Master/tlpkg/tlperl/lib/DBM/Deep/File.pm new file mode 100755 index 00000000000..13342d82465 --- /dev/null +++ b/Master/tlpkg/tlperl/lib/DBM/Deep/File.pm @@ -0,0 +1,277 @@ +package DBM::Deep::File; + +use 5.006_000; + +use strict; +use warnings; + +our $VERSION = q(1.0013); + +use Fcntl qw( :DEFAULT :flock :seek ); + +use constant DEBUG => 0; + +sub new { + my $class = shift; + my ($args) = @_; + + my $self = bless { + autobless => 1, + autoflush => 1, + end => 0, + fh => undef, + file => undef, + file_offset => 0, + locking => 1, + locked => 0, +#XXX Migrate this to the engine, where it really belongs. + filter_store_key => undef, + filter_store_value => undef, + filter_fetch_key => undef, + filter_fetch_value => undef, + }, $class; + + # Grab the parameters we want to use + foreach my $param ( keys %$self ) { + next unless exists $args->{$param}; + $self->{$param} = $args->{$param}; + } + + if ( $self->{fh} && !$self->{file_offset} ) { + $self->{file_offset} = tell( $self->{fh} ); + } + + $self->open unless $self->{fh}; + + return $self; +} + +sub open { + my $self = shift; + + # Adding O_BINARY should remove the need for the binmode below. However, + # I'm not going to remove it because I don't have the Win32 chops to be + # absolutely certain everything will be ok. + my $flags = O_CREAT | O_BINARY; + + if ( !-e $self->{file} || -w _ ) { + $flags |= O_RDWR; + } + else { + $flags |= O_RDONLY; + } + + my $fh; + sysopen( $fh, $self->{file}, $flags ) + or die "DBM::Deep: Cannot sysopen file '$self->{file}': $!\n"; + $self->{fh} = $fh; + + # Even though we use O_BINARY, better be safe than sorry. + binmode $fh; + + if ($self->{autoflush}) { + my $old = select $fh; + $|=1; + select $old; + } + + return 1; +} + +sub close { + my $self = shift; + + if ( $self->{fh} ) { + close $self->{fh}; + $self->{fh} = undef; + } + + return 1; +} + +sub set_inode { + my $self = shift; + + unless ( defined $self->{inode} ) { + my @stats = stat($self->{fh}); + $self->{inode} = $stats[1]; + $self->{end} = $stats[7]; + } + + return 1; +} + +sub print_at { + my $self = shift; + my $loc = shift; + + local ($/,$\); + + my $fh = $self->{fh}; + if ( defined $loc ) { + seek( $fh, $loc + $self->{file_offset}, SEEK_SET ); + } + + if ( DEBUG ) { + my $caller = join ':', (caller)[0,2]; + my $len = length( join '', @_ ); + warn "($caller) print_at( " . (defined $loc ? $loc : '<undef>') . ", $len )\n"; + } + + print( $fh @_ ) or die "Internal Error (print_at($loc)): $!\n"; + + return 1; +} + +sub read_at { + my $self = shift; + my ($loc, $size) = @_; + + local ($/,$\); + + my $fh = $self->{fh}; + if ( defined $loc ) { + seek( $fh, $loc + $self->{file_offset}, SEEK_SET ); + } + + if ( DEBUG ) { + my $caller = join ':', (caller)[0,2]; + warn "($caller) read_at( " . (defined $loc ? $loc : '<undef>') . ", $size )\n"; + } + + my $buffer; + read( $fh, $buffer, $size); + + return $buffer; +} + +sub DESTROY { + my $self = shift; + return unless $self; + + $self->close; + + return; +} + +sub request_space { + my $self = shift; + my ($size) = @_; + + #XXX Do I need to reset $self->{end} here? I need a testcase + my $loc = $self->{end}; + $self->{end} += $size; + + return $loc; +} + +## +# If db locking is set, flock() the db file. If called multiple +# times before unlock(), then the same number of unlocks() must +# be called before the lock is released. +## +sub lock { + my $self = shift; + my ($obj, $type) = @_; + + $type = LOCK_EX unless defined $type; + + #XXX This is a temporary fix for Win32 and autovivification. It + # needs to improve somehow. -RobK, 2008-03-09 + if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) { + $type = LOCK_EX; + } + + if (!defined($self->{fh})) { return; } + + #XXX This either needs to allow for upgrading a shared lock to an + # exclusive lock or something else with autovivification. + # -RobK, 2008-03-09 + if ($self->{locking}) { + if (!$self->{locked}) { + flock($self->{fh}, $type); + + # refresh end counter in case file has changed size + my @stats = stat($self->{fh}); + $self->{end} = $stats[7]; + + # double-check file inode, in case another process + # has optimize()d our file while we were waiting. + if (defined($self->{inode}) && $stats[1] != $self->{inode}) { + $self->close; + $self->open; + + #XXX This needs work + $obj->{engine}->setup_fh( $obj ); + + flock($self->{fh}, $type); # re-lock + + # This may not be necessary after re-opening + $self->{end} = (stat($self->{fh}))[7]; # re-end + } + } + $self->{locked}++; + + return 1; + } + + return; +} + +## +# If db locking is set, unlock the db file. See note in lock() +# regarding calling lock() multiple times. +## +sub unlock { + my $self = shift; + + if (!defined($self->{fh})) { return; } + + if ($self->{locking} && $self->{locked} > 0) { + $self->{locked}--; + if (!$self->{locked}) { flock($self->{fh}, LOCK_UN); } + + return 1; + } + + return; +} + +sub flush { + my $self = shift; + + # Flush the filehandle + my $old_fh = select $self->{fh}; + my $old_af = $|; $| = 1; $| = $old_af; + select $old_fh; + + return 1; +} + +# Taken from http://www.perlmonks.org/?node_id=691054 +sub is_writable { + my $self = shift; + + my $fh = $self->{fh}; + return unless defined $fh; + return unless defined fileno $fh; + local $\ = ''; # just in case + no warnings; # temporarily disable warnings + local $^W; # temporarily disable warnings + return print $fh ''; +} + +sub copy_stats { + my $self = shift; + my ($temp_filename) = @_; + + my @stats = stat( $self->{fh} ); + my $perms = $stats[2] & 07777; + my $uid = $stats[4]; + my $gid = $stats[5]; + chown( $uid, $gid, $temp_filename ); + chmod( $perms, $temp_filename ); +} + +1; +__END__ diff --git a/Master/tlpkg/tlperl/lib/DBM/Deep/Hash.pm b/Master/tlpkg/tlperl/lib/DBM/Deep/Hash.pm new file mode 100755 index 00000000000..6db4e218f3f --- /dev/null +++ b/Master/tlpkg/tlperl/lib/DBM/Deep/Hash.pm @@ -0,0 +1,136 @@ +package DBM::Deep::Hash; + +use 5.006_000; + +use strict; +use warnings; + +our $VERSION = q(1.0013); + +use base 'DBM::Deep'; + +sub _get_self { + eval { local $SIG{'__DIE__'}; tied( %{$_[0]} ) } || $_[0] +} + +sub _repr { return {} } + +sub TIEHASH { + ## + # Tied hash constructor method, called by Perl's tie() function. + ## + my $class = shift; + my $args = $class->_get_args( @_ ); + + $args->{type} = $class->TYPE_HASH; + + return $class->_init($args); +} + +sub FETCH { + my $self = shift->_get_self; + DBM::Deep->_throw_error( "Cannot use an undefined hash key." ) unless defined $_[0]; + my $key = ($self->_storage->{filter_store_key}) + ? $self->_storage->{filter_store_key}->($_[0]) + : $_[0]; + + return $self->SUPER::FETCH( $key, $_[0] ); +} + +sub STORE { + my $self = shift->_get_self; + DBM::Deep->_throw_error( "Cannot use an undefined hash key." ) unless defined $_[0]; + my $key = ($self->_storage->{filter_store_key}) + ? $self->_storage->{filter_store_key}->($_[0]) + : $_[0]; + my $value = $_[1]; + + return $self->SUPER::STORE( $key, $value, $_[0] ); +} + +sub EXISTS { + my $self = shift->_get_self; + DBM::Deep->_throw_error( "Cannot use an undefined hash key." ) unless defined $_[0]; + my $key = ($self->_storage->{filter_store_key}) + ? $self->_storage->{filter_store_key}->($_[0]) + : $_[0]; + + return $self->SUPER::EXISTS( $key ); +} + +sub DELETE { + my $self = shift->_get_self; + DBM::Deep->_throw_error( "Cannot use an undefined hash key." ) unless defined $_[0]; + my $key = ($self->_storage->{filter_store_key}) + ? $self->_storage->{filter_store_key}->($_[0]) + : $_[0]; + + return $self->SUPER::DELETE( $key, $_[0] ); +} + +sub FIRSTKEY { + ## + # Locate and return first key (in no particular order) + ## + my $self = shift->_get_self; + + ## + # Request shared lock for reading + ## + $self->lock( $self->LOCK_SH ); + + my $result = $self->_engine->get_next_key( $self ); + + $self->unlock(); + + return ($result && $self->_storage->{filter_fetch_key}) + ? $self->_storage->{filter_fetch_key}->($result) + : $result; +} + +sub NEXTKEY { + ## + # Return next key (in no particular order), given previous one + ## + my $self = shift->_get_self; + + my $prev_key = ($self->_storage->{filter_store_key}) + ? $self->_storage->{filter_store_key}->($_[0]) + : $_[0]; + + ## + # Request shared lock for reading + ## + $self->lock( $self->LOCK_SH ); + + my $result = $self->_engine->get_next_key( $self, $prev_key ); + + $self->unlock(); + + return ($result && $self->_storage->{filter_fetch_key}) + ? $self->_storage->{filter_fetch_key}->($result) + : $result; +} + +## +# Public method aliases +## +sub first_key { (shift)->FIRSTKEY(@_) } +sub next_key { (shift)->NEXTKEY(@_) } + +sub _copy_node { + my $self = shift; + my ($db_temp) = @_; + + my $key = $self->first_key(); + while ($key) { + my $value = $self->get($key); + $self->_copy_value( \$db_temp->{$key}, $value ); + $key = $self->next_key($key); + } + + return 1; +} + +1; +__END__ diff --git a/Master/tlpkg/tlperl/lib/DBM/Deep/Internals.pod b/Master/tlpkg/tlperl/lib/DBM/Deep/Internals.pod new file mode 100755 index 00000000000..132bc9eff17 --- /dev/null +++ b/Master/tlpkg/tlperl/lib/DBM/Deep/Internals.pod @@ -0,0 +1,281 @@ +=head1 NAME + +DBM::Deep::Internals + +=head1 DESCRIPTION + +B<NOTE>: This document is out-of-date. It describes an intermediate file +format used during the development from 0.983 to 1.0000. It will be rewritten +soon. + +This is a document describing the internal workings of L<DBM::Deep>. It is +not necessary to read this document if you only intend to be a user. This +document is intended for people who either want a deeper understanding of +specifics of how L<DBM::Deep> works or who wish to help program +L<DBM::Deep>. + +=head1 CLASS LAYOUT + +L<DBM::Deep> is broken up into five classes in three inheritance hierarchies. + +=over 4 + +=item * + +L<DBM::Deep> is the parent of L<DBM::Deep::Array> and L<DBM::Deep::Hash>. +These classes form the immediate interface to the outside world. They are the +classes that provide the TIE mechanisms as well as the OO methods. + +=item * + +L<DBM::Deep::Engine> is the layer that deals with the mechanics of reading +and writing to the file. This is where the logic of the file layout is +handled. + +=item * + +L<DBM::Deep::File> is the layer that deals with the physical file. As a +singleton that every other object has a reference to, it also provides a place +to handle datastructure-wide items, such as transactions. + +=back + +=head1 FILE LAYOUT + +DBM::Deep uses a tagged file layout. Every section has a tag, a size, and then +the data. + +=head2 File header + +=over 4 + +=item * File Signature + +The first four bytes are 'DPDB' in network byte order, signifying that this is +a DBM::Deep file. + +=item * File tag/size + +This is the tagging of the file header. The file used by versions prior to +1.00 had a different fifth byte, allowing the difference to the determined. + +=item * Version + +This is four bytes containing the file version. This lets the file format change over time. + +=item * Constants + +These are the file-wide constants that determine how the file is laid out. +They can only be set upon file creation. + +=item * Transaction information + +The current running transactions are stored here, as is the next transaction +ID. + +=item * Freespace information + +Pointers into the next free sectors of the various sector sizes (Index, +Bucketlist, and Data) are stored here. + +=back + +=head2 Index + +The Index parts can be tagged either as Hash, Array, or Index. The latter +is if there was a reindexing due to a bucketlist growing too large. The others +are the root index for their respective datatypes. The index consists of a +tag, a size, and then 256 sections containing file locations. Each section +corresponds to each value representable in a byte. + +The index is used as follows - whenever a hashed key is being looked up, the +first byte is used to determine which location to go to from the root index. +Then, if that's also an index, the second byte is used, and so forth until a +bucketlist is found. + +=head2 Bucketlist + +This is the part that contains the link to the data section. A bucketlist +defaults to being 16 buckets long (modifiable by the I<max_buckets> +parameter used when creating a new file). Each bucket contains an MD5 and a +location of the appropriate key section. + +=head2 Key area + +This is the part that handles transactional awareness. There are +I<max_buckets> sections. Each section contains the location to the data +section, a transaction ID, and whether that transaction considers this key to +be deleted or not. + +=head2 Data area + +This is the part that actual stores the key, value, and class (if +appropriate). The layout is: + +=over 4 + +=item * tag + +=item * length of the value + +=item * the actual value + +=item * keylength + +=item * the actual key + +=item * a byte indicating if this value has a classname + +=item * the classname (if one is there) + +=back + +The key is stored after the value because the value is requested more often +than the key. + +=head1 PERFORMANCE + +L<DBM::Deep> is written completely in Perl. It also is a multi-process DBM +that uses the datafile as a method of synchronizing between multiple +processes. This is unlike most RDBMSes like MySQL and Oracle. Furthermore, +unlike all RDBMSes, L<DBM::Deep> stores both the data and the structure of +that data as it would appear in a Perl program. + +=head2 CPU + +DBM::Deep attempts to be CPU-light. As it stores all the data on disk, +DBM::Deep is I/O-bound, not CPU-bound. + +=head2 RAM + +DBM::Deep uses extremely little RAM relative to the amount of data you can +access. You can iterate through a million keys (using C<each()>) without +increasing your memeory usage at all. + +=head2 DISK + +DBM::Deep is I/O-bound, pure and simple. The faster your disk, the faster +DBM::Deep will be. Currently, when performing C<my $x = $db-E<gt>{foo}>, there +are a minimum of 4 seeks and 1332 + N bytes read (where N is the length of your +data). (All values assume a medium filesize.) The actions taken are: + +=over 4 + +=item 1 Lock the file + +=item 1 Perform a stat() to determine if the inode has changed + +=item 1 Go to the primary index for the $db (1 seek) + +=item 1 Read the tag/size of the primary index (5 bytes) + +=item 1 Read the body of the primary index (1024 bytes) + +=item 1 Go to the bucketlist for this MD5 (1 seek) + +=item 1 Read the tag/size of the bucketlist (5 bytes) + +=item 1 Read the body of the bucketlist (144 bytes) + +=item 1 Go to the keys location for this MD5 (1 seek) + +=item 1 Read the tag/size of the keys section (5 bytes) + +=item 1 Read the body of the keys location (144 bytes) + +=item 1 Go to the data section that corresponds to this transaction ID. (1 seek) + +=item 1 Read the tag/size of the data section (5 bytes) + +=item 1 Read the value for this data (N bytes) + +=item 1 Unlock the file + +=back + +Every additional level of indexing (if there are enough keys) requires an +additional seek and the reading of 1029 additional bytes. If the value is +blessed, an additional 1 seek and 9 + M bytes are read (where M is the length +of the classname). + +Arrays are (currently) even worse because they're considered "funny hashes" +with the length stored as just another key. This means that if you do any sort +of lookup with a negative index, this entire process is performed twice - once +for the length and once for the value. + +=head1 ACTUAL TESTS + +=head2 SPEED + +Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as +the almighty I<BerkeleyDB>. But it makes up for it in features like true +multi-level hash/array support, and cross-platform FTPable files. Even so, +DBM::Deep is still pretty fast, and the speed stays fairly consistent, even +with huge databases. Here is some test data: + + Adding 1,000,000 keys to new DB file... + + At 100 keys, avg. speed is 2,703 keys/sec + At 200 keys, avg. speed is 2,642 keys/sec + At 300 keys, avg. speed is 2,598 keys/sec + At 400 keys, avg. speed is 2,578 keys/sec + At 500 keys, avg. speed is 2,722 keys/sec + At 600 keys, avg. speed is 2,628 keys/sec + At 700 keys, avg. speed is 2,700 keys/sec + At 800 keys, avg. speed is 2,607 keys/sec + At 900 keys, avg. speed is 2,190 keys/sec + At 1,000 keys, avg. speed is 2,570 keys/sec + At 2,000 keys, avg. speed is 2,417 keys/sec + At 3,000 keys, avg. speed is 1,982 keys/sec + At 4,000 keys, avg. speed is 1,568 keys/sec + At 5,000 keys, avg. speed is 1,533 keys/sec + At 6,000 keys, avg. speed is 1,787 keys/sec + At 7,000 keys, avg. speed is 1,977 keys/sec + At 8,000 keys, avg. speed is 2,028 keys/sec + At 9,000 keys, avg. speed is 2,077 keys/sec + At 10,000 keys, avg. speed is 2,031 keys/sec + At 20,000 keys, avg. speed is 1,970 keys/sec + At 30,000 keys, avg. speed is 2,050 keys/sec + At 40,000 keys, avg. speed is 2,073 keys/sec + At 50,000 keys, avg. speed is 1,973 keys/sec + At 60,000 keys, avg. speed is 1,914 keys/sec + At 70,000 keys, avg. speed is 2,091 keys/sec + At 80,000 keys, avg. speed is 2,103 keys/sec + At 90,000 keys, avg. speed is 1,886 keys/sec + At 100,000 keys, avg. speed is 1,970 keys/sec + At 200,000 keys, avg. speed is 2,053 keys/sec + At 300,000 keys, avg. speed is 1,697 keys/sec + At 400,000 keys, avg. speed is 1,838 keys/sec + At 500,000 keys, avg. speed is 1,941 keys/sec + At 600,000 keys, avg. speed is 1,930 keys/sec + At 700,000 keys, avg. speed is 1,735 keys/sec + At 800,000 keys, avg. speed is 1,795 keys/sec + At 900,000 keys, avg. speed is 1,221 keys/sec + At 1,000,000 keys, avg. speed is 1,077 keys/sec + +This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl +5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and +values were between 6 - 12 chars in length. The DB file ended up at 210MB. +Run time was 12 min 3 sec. + +=head2 MEMORY USAGE + +One of the great things about L<DBM::Deep> is that it uses very little memory. +Even with huge databases (1,000,000+ keys) you will not see much increased +memory on your process. L<DBM::Deep> relies solely on the filesystem for storing +and fetching data. Here is output from I<top> before even opening a database +handle: + + PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND + 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl + +Basically the process is taking 2,716K of memory. And here is the same +process after storing and fetching 1,000,000 keys: + + PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND + 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl + +Notice the memory usage increased by only 56K. Test was performed on a 700mHz +x86 box running Linux RedHat 7.2 & Perl 5.6.1. + +=cut |