| 1 |
package Class::Tangram; |
| 2 |
|
| 3 |
# Copyright (c) 2001 Sam Vilain. All rights reserved. This program is |
| 4 |
# free software; you can redistribute it and/or modify it under the |
| 5 |
# same terms as Perl itself. |
| 6 |
|
| 7 |
# Some modifications |
| 8 |
# $Id: Tangram.pm,v 1.1 2002/07/27 00:20:12 cvsjoko Exp $ |
| 9 |
# Copyright © 2001 Micro Sharp Technologies, Inc., Vancouver, WA, USA |
| 10 |
# Author: Karl M. Hegbloom <karlheg@microsharp.com> |
| 11 |
# Perl Artistic Licence. |
| 12 |
|
| 13 |
=head1 NAME |
| 14 |
|
| 15 |
Class::Tangram - create constructors, accessor and update methods for |
| 16 |
objects from a Tangram-compatible object specification. |
| 17 |
|
| 18 |
=head1 SYNOPSIS |
| 19 |
|
| 20 |
package Orange; |
| 21 |
|
| 22 |
use base qw(Class::Tangram); |
| 23 |
use vars qw($schema); |
| 24 |
use Tangram::Ref; |
| 25 |
|
| 26 |
# define the schema (ie, allowed attributes) of this object. See the |
| 27 |
# Tangram::Schema man page for more information on the syntax here. |
| 28 |
$schema = { |
| 29 |
table => "oranges", |
| 30 |
|
| 31 |
fields => { |
| 32 |
int => { |
| 33 |
juiciness => undef, |
| 34 |
segments => { |
| 35 |
# here is a new one - this code reference is called |
| 36 |
# when this attribute is set; it should die() on |
| 37 |
# error, as it is wrapped in an eval { } block |
| 38 |
check_func => sub { |
| 39 |
die "too many segments" |
| 40 |
if ($ {$_[0]} > 30); |
| 41 |
}, |
| 42 |
# the default for this attribute. |
| 43 |
init_default => 7, |
| 44 |
}, |
| 45 |
}, |
| 46 |
ref => { |
| 47 |
grower => undef, |
| 48 |
}, |
| 49 |
}, |
| 50 |
}; |
| 51 |
Class::Tangram::import_schema("Orange"); |
| 52 |
|
| 53 |
package Project; |
| 54 |
# here's where we build the individual object schemas into a |
| 55 |
# Tangram::Schema object, which the Tangram::Storage class uses to |
| 56 |
# know which tables and columns to find objects. |
| 57 |
use Tangram::Schema; |
| 58 |
|
| 59 |
my $dbschema = Tangram::Schema->new |
| 60 |
({ classes => [ 'Orange' => $Orange::schema ]}); |
| 61 |
|
| 62 |
sub schema { $dbschema }; |
| 63 |
|
| 64 |
package main; |
| 65 |
|
| 66 |
# See Tangram::Relational for instructions on using "deploy" to |
| 67 |
# create the database this connects to. You only have to do this if |
| 68 |
# you want to write the objects to a database. |
| 69 |
use Tangram::Relational; |
| 70 |
my ($dsn, $u, $p); |
| 71 |
my $storage = Tangram::Relational->connect(Project->schema, |
| 72 |
$dsn, $u, $p); |
| 73 |
|
| 74 |
# OK |
| 75 |
my $orange = Orange->new(juiciness => 8); |
| 76 |
my $juiciness = $orange->juiciness; # returns 8 |
| 77 |
|
| 78 |
# a "ref" must be set to a blessed object |
| 79 |
my $grower = bless { name => "Joe" }, "Farmer"; |
| 80 |
$orange->set_grower ($grower); |
| 81 |
|
| 82 |
# these are all illegal |
| 83 |
eval { $orange->set_juiciness ("Yum"); }; print $@; |
| 84 |
eval { $orange->set_segments (31); }; print $@; |
| 85 |
eval { $orange->set_grower ("Mr. Nice"); }; print $@; |
| 86 |
|
| 87 |
# if you prefer |
| 88 |
$orange->get( "juiciness" ); |
| 89 |
$orange->set( juiciness => 123 ); |
| 90 |
|
| 91 |
=head1 DESCRIPTION |
| 92 |
|
| 93 |
Class::Tangram is a base class originally intended for use with |
| 94 |
Tangram objects, that gives you free constructors, access methods, |
| 95 |
update methods, and a destructor that should help in breaking circular |
| 96 |
references for you. Type checking is achieved by parsing the schema |
| 97 |
for the object, which is contained within the object class in an |
| 98 |
exported variable C<$schema>. |
| 99 |
|
| 100 |
After writing this I found that it was useful for merely adding type |
| 101 |
checking and validation to arbitrary objects. There are several |
| 102 |
modules on CPAN to do that already, but many don't have finely grained |
| 103 |
type checking, and none of them integrated with Tangram. |
| 104 |
|
| 105 |
=cut |
| 106 |
|
| 107 |
use strict; |
| 108 |
use Carp qw(croak cluck); |
| 109 |
|
| 110 |
use vars qw($AUTOLOAD $VERSION); |
| 111 |
$VERSION = "1.04"; |
| 112 |
|
| 113 |
local $AUTOLOAD; |
| 114 |
|
| 115 |
# $types{$class}->{$attribute} is the tangram type of each attribute |
| 116 |
my (%types); |
| 117 |
|
| 118 |
# $check{$class}->{$attribute}->($value) is a function that will die |
| 119 |
# if $value is not alright, see check_X functions |
| 120 |
my (%check); |
| 121 |
|
| 122 |
# Destructors for each attribute. They are called as |
| 123 |
# $cleaners{$class}->{$attribute}->($self, $attribute); |
| 124 |
my (%cleaners); |
| 125 |
|
| 126 |
# init_default values for each attribute. These could be hash refs, |
| 127 |
# array refs, code refs, or simple scalars. They will be stored as |
| 128 |
# $init_defaults{$class}->{$attribute} |
| 129 |
my (%init_defaults); |
| 130 |
|
| 131 |
# if a class is abstract, complain if one is constructed. |
| 132 |
my (%abstract); |
| 133 |
|
| 134 |
=head1 METHODS |
| 135 |
|
| 136 |
=over 4 |
| 137 |
|
| 138 |
=item Class-E<gt>new (attribute1 =E<gt> value, attribute2 =E<gt> value) |
| 139 |
|
| 140 |
sets up a new object of type Class, with attributes set to the values |
| 141 |
supplied. |
| 142 |
|
| 143 |
Can also be used as an object method, in which case it returns a |
| 144 |
B<copy> of the object, without any deep copying. |
| 145 |
|
| 146 |
=cut |
| 147 |
|
| 148 |
sub new ($@) |
| 149 |
{ |
| 150 |
my $invocant = shift; |
| 151 |
my $class = ref $invocant || $invocant; |
| 152 |
|
| 153 |
my @values = @_; |
| 154 |
|
| 155 |
# Setup the object |
| 156 |
my $self = { }; |
| 157 |
bless $self, $class; |
| 158 |
|
| 159 |
exists $check{$class} or import_schema($class); |
| 160 |
|
| 161 |
croak "Attempt to instantiate an abstract type" |
| 162 |
if ($abstract{$class}); |
| 163 |
|
| 164 |
if ($invocant ne $class) |
| 165 |
{ |
| 166 |
# The copy constructor; this could be better :) |
| 167 |
# this has the side effect of much auto-vivification. |
| 168 |
%$self = %$invocant; |
| 169 |
$self->set (@values); # override with @values |
| 170 |
} |
| 171 |
else |
| 172 |
{ |
| 173 |
$self->set (@values); # start with @values |
| 174 |
|
| 175 |
# now fill in fields that have defaults |
| 176 |
for my $attribute (keys %{$init_defaults{$class}}) { |
| 177 |
|
| 178 |
next if (exists $self->{$attribute}); |
| 179 |
|
| 180 |
my $default = $init_defaults{$class}->{$attribute} |
| 181 |
unless tied $init_defaults{$class}->{$attribute}; |
| 182 |
|
| 183 |
if (ref $default eq "CODE") { |
| 184 |
# sub { }, attribute gets return value |
| 185 |
$self->{$attribute} |
| 186 |
= $init_defaults{$class}->{$attribute}->(); |
| 187 |
|
| 188 |
} elsif (ref $default eq "HASH") { |
| 189 |
# hash ref, copy hash |
| 190 |
$self->{$attribute} |
| 191 |
= { %{ $init_defaults{$class}->{$attribute} } }; |
| 192 |
|
| 193 |
} elsif (ref $default eq "ARRAY") { |
| 194 |
# array ref, copy array |
| 195 |
$self->{$attribute} |
| 196 |
= [ @{ $init_defaults{$class}->{$attribute} } ]; |
| 197 |
|
| 198 |
} else { |
| 199 |
# something else, an object or a scalar |
| 200 |
$self->{$attribute} |
| 201 |
= $init_defaults{$class}->{$attribute}; |
| 202 |
} |
| 203 |
} |
| 204 |
} |
| 205 |
return $self; |
| 206 |
} |
| 207 |
|
| 208 |
=item $instance->set(attribute => $value, ...) |
| 209 |
|
| 210 |
Sets the attributes of the given instance to the given values. croaks |
| 211 |
if there is a problem with the values. |
| 212 |
|
| 213 |
=cut |
| 214 |
|
| 215 |
sub set($@) { |
| 216 |
my ($self, @values) = (@_); |
| 217 |
|
| 218 |
# yes, this is a lot to do. yes, it's slow. But I'm fairly |
| 219 |
# certain that this could be handled efficiently if it were to be |
| 220 |
# moved inside the Perl interpreter or an XS module |
| 221 |
$self->isa("Class::Tangram") or croak "type mismatch"; |
| 222 |
my $class = ref $self; |
| 223 |
exists $check{$class} or import_schema($class); |
| 224 |
|
| 225 |
while (my ($name, $value) = splice @values, 0, 2) { |
| 226 |
croak "attempt to set an illegal field $name in a $class" |
| 227 |
if (!defined $check{$class}->{$name}); |
| 228 |
|
| 229 |
#local $@; |
| 230 |
|
| 231 |
# these handlers die on failure |
| 232 |
eval { $check{$class}->{$name}->(\$value) }; |
| 233 |
$@ && croak ("value failed type check - ${class}->{$name}, " |
| 234 |
."\"$value\" ($@)"); |
| 235 |
|
| 236 |
#should be ok now |
| 237 |
$self->{$name} = $value; |
| 238 |
} |
| 239 |
} |
| 240 |
|
| 241 |
=item $instance->get($attribute) |
| 242 |
|
| 243 |
Gets the value of $attribute. If the attribute in question is a set, |
| 244 |
and this method is called in list context, then it returns the MEMBERS |
| 245 |
of the set (if called in scalar context, it returns the Set::Object |
| 246 |
container). |
| 247 |
|
| 248 |
=cut |
| 249 |
|
| 250 |
sub get($$) { |
| 251 |
my ($self, $field) = (@_); |
| 252 |
$self->isa("Class::Tangram") or croak "type mismatch"; |
| 253 |
my $class = ref $self; |
| 254 |
exists $check{$class} or import_schema($class); |
| 255 |
croak "attempt to read an illegal field $field in a $class" |
| 256 |
if (!defined $check{$class}->{$field}); |
| 257 |
|
| 258 |
if ($types{$class}->{$field} =~ m/^i?set$/o) { |
| 259 |
if (!defined $self->{$field}) { |
| 260 |
$self->{$field} = Set::Object->new(); |
| 261 |
} |
| 262 |
if (wantarray) { |
| 263 |
return $self->{$field}->members; |
| 264 |
} |
| 265 |
} |
| 266 |
|
| 267 |
return $self->{$field}; |
| 268 |
} |
| 269 |
|
| 270 |
=item $instance->attribute($value) |
| 271 |
|
| 272 |
If $value is not given, then |
| 273 |
this is equivalent to $instance->get("attribute") |
| 274 |
|
| 275 |
If $value is given, then this is equivalent to |
| 276 |
$instance->set("attribute", $value). This usage issues a warning; you |
| 277 |
should change your code to use the set_attribute syntax for better |
| 278 |
readability. |
| 279 |
|
| 280 |
=item $instance->get_attribute |
| 281 |
|
| 282 |
=item $instance->set_attribute($value) |
| 283 |
|
| 284 |
Equivalent to $instance->get("attribute") and $instance->set(attribute |
| 285 |
=> $value), respectively. |
| 286 |
|
| 287 |
=item $instance->attribute_includes(@objects) |
| 288 |
|
| 289 |
=item $instance->attribute_insert(@objects) |
| 290 |
|
| 291 |
=item $instance->attribute_size |
| 292 |
|
| 293 |
=item $instance->attribute_clear |
| 294 |
|
| 295 |
=item $instance->attribute_remove(@objects) |
| 296 |
|
| 297 |
Equivalent to calling $instance->attribute->includes(@objects), etc. |
| 298 |
This only works if the attribute in question is a Set::Object. |
| 299 |
|
| 300 |
=cut |
| 301 |
|
| 302 |
sub AUTOLOAD ($;$) { |
| 303 |
my ($self, $value) = (@_); |
| 304 |
$self->isa("Class::Tangram") or croak "type mismatch"; |
| 305 |
|
| 306 |
my $class = ref $self; |
| 307 |
$AUTOLOAD =~ s/.*://; |
| 308 |
if ($AUTOLOAD =~ m/^(set_|get_)?([^:]+)$/ |
| 309 |
and defined $types{$class}->{$2}) { |
| 310 |
|
| 311 |
# perl sucks at this type of test |
| 312 |
if ((defined $1 and $1 eq "set_") |
| 313 |
or (!defined $1 and defined $value)) { |
| 314 |
|
| 315 |
if ($^W && !defined $1) { |
| 316 |
cluck("The OO police say change your call to " |
| 317 |
."\$obj->set_$2"); |
| 318 |
} |
| 319 |
return set($self, $2, $value); |
| 320 |
} else { |
| 321 |
return get($self, $2); |
| 322 |
} |
| 323 |
} elsif (my ($attr, $method) = |
| 324 |
($AUTOLOAD =~ m/^(.*)_(includes|insert| |
| 325 |
size|clear|remove)$/x) |
| 326 |
and $types{$class}->{$1} =~ m/^i?set$/) { |
| 327 |
return get($self, $attr)->$method(@_[1..$#_]); |
| 328 |
} else { |
| 329 |
croak("unknown method/attribute ${class}->$AUTOLOAD called"); |
| 330 |
} |
| 331 |
} |
| 332 |
|
| 333 |
=item $instance->getset($attribute, $value) |
| 334 |
|
| 335 |
If you're replacing the AUTOLOAD function in your Class::Tangram |
| 336 |
derived class, but would still like to use the behaviour for one or |
| 337 |
two fields, then you can define functions for them to fall through to |
| 338 |
the Class::Tangram method, like so: |
| 339 |
|
| 340 |
sub attribute { $_[0]->SUPER::getset("attribute", $_[1]) } |
| 341 |
|
| 342 |
=cut |
| 343 |
|
| 344 |
sub getset($$;$) { |
| 345 |
my ($self, $attr, $value) = (@_); |
| 346 |
$self->isa("Class::Tangram") or croak "type mismatch"; |
| 347 |
|
| 348 |
if (defined $value) { |
| 349 |
return set($self, $attr, $value); |
| 350 |
} else { |
| 351 |
return get($self, $attr); |
| 352 |
} |
| 353 |
|
| 354 |
} |
| 355 |
|
| 356 |
=item check_X (\$value) |
| 357 |
|
| 358 |
This series of functions checks that $value is of the type X, and |
| 359 |
within applicable bounds. If there is a problem, then it will croak() |
| 360 |
the error. These functions are not called from the code, but by the |
| 361 |
set() method on a particular attribute. |
| 362 |
|
| 363 |
Available functions are: |
| 364 |
|
| 365 |
check_string - checks that the supplied value is less |
| 366 |
than 255 characters long. |
| 367 |
|
| 368 |
=cut |
| 369 |
|
| 370 |
sub check_string { |
| 371 |
croak "string ${$_[0]} too long" |
| 372 |
if (length ${$_[0]} > 255); |
| 373 |
} |
| 374 |
|
| 375 |
=pod |
| 376 |
|
| 377 |
check_int - checks that the value is a (possibly |
| 378 |
signed) integer |
| 379 |
|
| 380 |
=cut |
| 381 |
|
| 382 |
my $int_re = qr/^-?\d+$/; |
| 383 |
sub check_int { |
| 384 |
croak "not an int" |
| 385 |
if (${$_[0]} !~ m/$int_re/ms); |
| 386 |
} |
| 387 |
|
| 388 |
=pod |
| 389 |
|
| 390 |
check_real - checks that the value is a real number |
| 391 |
(m/^\d*(\.\d*)?(e\d*)?$/) |
| 392 |
|
| 393 |
=cut |
| 394 |
|
| 395 |
my $real_re = qr/^-?\d*(\.\d*)?(e-?\d*)?$/; |
| 396 |
sub check_real { |
| 397 |
croak "not a real" |
| 398 |
if (${$_[0]} !~ m/$real_re/ms); |
| 399 |
} |
| 400 |
|
| 401 |
=pod |
| 402 |
|
| 403 |
check_obj - checks that the supplied variable is a |
| 404 |
reference to a blessed object |
| 405 |
|
| 406 |
=cut |
| 407 |
|
| 408 |
# this pattern matches a regular reference |
| 409 |
my $obj_re = qr/^(?:HASH|ARRAY|SCALAR)?$/; |
| 410 |
sub check_obj { |
| 411 |
croak "not an object reference" |
| 412 |
if ((ref ${$_[0]}) =~ m/$obj_re/); |
| 413 |
} |
| 414 |
|
| 415 |
=pod |
| 416 |
|
| 417 |
check_flat_array |
| 418 |
- checks that $value is a ref ARRAY |
| 419 |
|
| 420 |
=cut |
| 421 |
|
| 422 |
sub check_flat_array { |
| 423 |
croak "not a flat array" |
| 424 |
if (ref ${$_[0]} ne "ARRAY"); |
| 425 |
} |
| 426 |
|
| 427 |
=pod |
| 428 |
|
| 429 |
check_array - checks that $value is a ref ARRAY, and that |
| 430 |
each element in the array is a reference to |
| 431 |
a blessed object. |
| 432 |
|
| 433 |
=cut |
| 434 |
|
| 435 |
sub check_array { |
| 436 |
croak "array attribute not passed an array ref" |
| 437 |
if (ref ${$_[0]} ne "ARRAY"); |
| 438 |
for my $a (@{${$_[0]}}) { |
| 439 |
croak "member in array not an object reference" |
| 440 |
if ((ref $a) =~ m/$obj_re/); |
| 441 |
} |
| 442 |
} |
| 443 |
|
| 444 |
=pod |
| 445 |
|
| 446 |
check_set - checks that $value->isa("Set::Object") |
| 447 |
|
| 448 |
=cut |
| 449 |
|
| 450 |
sub check_set { |
| 451 |
croak "set type not passed a Set::Object" |
| 452 |
unless (ref ${$_[0]} and ${$_[0]}->isa("Set::Object")); |
| 453 |
} |
| 454 |
|
| 455 |
=pod |
| 456 |
|
| 457 |
check_rawdatetime |
| 458 |
- checks that $value is of the form |
| 459 |
YYYY-MM-DD HH:MM:SS |
| 460 |
|
| 461 |
=cut |
| 462 |
|
| 463 |
# YYYY-MM-DD HH:MM:SS |
| 464 |
my $rawdatetime_re = qr/^\d{4}-\d{2}-\d{2}\s+\d{1,2}:\d{2}:\d{2}$/; |
| 465 |
sub check_rawdatetime { |
| 466 |
croak "invalid SQL rawdatetime" |
| 467 |
unless (${$_[0]} =~ m/$rawdatetime_re/); |
| 468 |
} |
| 469 |
|
| 470 |
=pod |
| 471 |
|
| 472 |
check_time |
| 473 |
- checks that $value is of the form |
| 474 |
HH:MM(:SS)? |
| 475 |
|
| 476 |
=cut |
| 477 |
|
| 478 |
my $time_re = qr/^\d{1,2}:\d{2}(?::\d{2})?$/; |
| 479 |
sub check_time { |
| 480 |
croak "invalid SQL time" |
| 481 |
unless (${$_[0]} =~ m/$time_re/); |
| 482 |
} |
| 483 |
|
| 484 |
=pod |
| 485 |
|
| 486 |
check_timestamp |
| 487 |
- checks that $value is of the form |
| 488 |
(YYYY-MM-DD )?HH:MM(:SS)? |
| 489 |
|
| 490 |
=cut |
| 491 |
|
| 492 |
my $timestamp_re = qr/^(?:\d{4}-\d{2}-\d{2}\s+)?\d{1,2}:\d{2}(?::\d{2})?$/; |
| 493 |
sub check_timestamp { |
| 494 |
croak "invalid SQL timestamp" |
| 495 |
unless (${$_[0]} =~ m/$timestamp_re/); |
| 496 |
} |
| 497 |
|
| 498 |
=pod |
| 499 |
|
| 500 |
check_flat_hash |
| 501 |
- checks that $value is a ref HASH |
| 502 |
|
| 503 |
=cut |
| 504 |
|
| 505 |
sub check_flat_hash { |
| 506 |
croak "not a hash" |
| 507 |
unless (ref ${$_[0]} eq "HASH"); |
| 508 |
while (my ($k, $v) = each %${$_[0]}) { |
| 509 |
croak "hash not flat" |
| 510 |
if (ref $k or ref $v); |
| 511 |
} |
| 512 |
} |
| 513 |
|
| 514 |
=pod |
| 515 |
|
| 516 |
check_hash - checks that $value is a ref HASH, that |
| 517 |
every key in the hash is a scalar, and that |
| 518 |
every value is a blessed object. |
| 519 |
|
| 520 |
=cut |
| 521 |
|
| 522 |
sub check_hash { |
| 523 |
croak "not a hash" |
| 524 |
unless (ref ${$_[0]} eq "HASH"); |
| 525 |
while (my ($k, $v) = each %${$_[0]}) { |
| 526 |
croak "hash key not flat" |
| 527 |
if (ref $k); |
| 528 |
croak "hash value not an object" |
| 529 |
if (ref $v !~ m/$obj_re/); |
| 530 |
} |
| 531 |
} |
| 532 |
|
| 533 |
=pod |
| 534 |
|
| 535 |
check_nothing - checks whether Australians like sport |
| 536 |
|
| 537 |
=cut |
| 538 |
|
| 539 |
sub check_nothing { } |
| 540 |
|
| 541 |
=item destroy_X ($instance, $attr) |
| 542 |
|
| 543 |
Similar story with the check_X series of functions, these are called |
| 544 |
during object destruction on every attribute that has a reference that |
| 545 |
might need breaking. Note: B<these functions all assume that |
| 546 |
attributes belonging to an object that is being destroyed may be |
| 547 |
destroyed also>. In other words, do not allow distinct objects to |
| 548 |
share Set::Object containers or hash references in their attributes, |
| 549 |
otherwise when one gets destroyed the others will lose their data. |
| 550 |
|
| 551 |
Available functions are: |
| 552 |
|
| 553 |
destroy_array - empties an array |
| 554 |
|
| 555 |
=cut |
| 556 |
|
| 557 |
sub destroy_array { |
| 558 |
my ($self, $attr) = (@_); |
| 559 |
my $t = tied $self->{$attr}; |
| 560 |
@{$self->{$attr}} = () unless ($t =~ m,Tangram::CollOnDemand,); |
| 561 |
delete $self->{$attr}; |
| 562 |
} |
| 563 |
|
| 564 |
=pod |
| 565 |
|
| 566 |
destroy_set - calls Set::Object::clear to clear the set |
| 567 |
|
| 568 |
=cut |
| 569 |
|
| 570 |
sub destroy_set { |
| 571 |
my ($self, $attr) = (@_); |
| 572 |
|
| 573 |
# warnings suck sometimes |
| 574 |
local $^W = 0; |
| 575 |
|
| 576 |
my $t = tied $self->{$attr}; |
| 577 |
return if ($t =~ m,Tangram::CollOnDemand,); |
| 578 |
if (ref $self->{$attr} eq "Set::Object") { |
| 579 |
$self->{$attr}->clear; |
| 580 |
} |
| 581 |
delete $self->{$attr}; |
| 582 |
} |
| 583 |
|
| 584 |
=pod |
| 585 |
|
| 586 |
destroy_hash - empties a hash |
| 587 |
|
| 588 |
=cut |
| 589 |
|
| 590 |
sub destroy_hash { |
| 591 |
my ($self, $attr) = (@_); |
| 592 |
my $t = tied $self->{$attr}; |
| 593 |
%{$self->{$attr}} = () unless ($t =~ m,Tangram::CollOnDemand,); |
| 594 |
delete $self->{$attr}; |
| 595 |
} |
| 596 |
|
| 597 |
=pod |
| 598 |
|
| 599 |
destroy_ref - destroys a reference. Contains a hack for |
| 600 |
Tangram so that if this ref is not loaded, |
| 601 |
it will not be autoloaded when it is |
| 602 |
attempted to be accessed. |
| 603 |
|
| 604 |
=cut |
| 605 |
|
| 606 |
sub destroy_ref { |
| 607 |
my ($self, $attr) = (@_); |
| 608 |
|
| 609 |
# warnings suck sometimes |
| 610 |
local $^W = 0; |
| 611 |
|
| 612 |
# the only reason I bother with all of this is that I experienced |
| 613 |
# Perl did not always call an object's destructor if you just used |
| 614 |
# delete. |
| 615 |
my $t = tied $self->{$attr}; |
| 616 |
if (defined $t and $t =~ m/OnDemand/) { |
| 617 |
delete $self->{$attr}; |
| 618 |
} else { |
| 619 |
my $ref = delete $self->{$attr}; |
| 620 |
} |
| 621 |
} |
| 622 |
|
| 623 |
=item parse_X ($attribute, { schema option }) |
| 624 |
|
| 625 |
Parses the schema option field, and returns one or two closures that |
| 626 |
act as a check_X and a destroy_X function for the attribute. |
| 627 |
|
| 628 |
This is currently a very ugly hack, parsing the SQL type definition of |
| 629 |
an object. But it was bloody handy in my case for hacking this in |
| 630 |
quickly. This is unmanagably unportable across databases. This |
| 631 |
should be replaced by primitives that go the other way, building the |
| 632 |
SQL type definition from a more abstract definition of the type. |
| 633 |
|
| 634 |
Available functions: |
| 635 |
|
| 636 |
parse_string - parses SQL types of: |
| 637 |
|
| 638 |
=cut |
| 639 |
|
| 640 |
sub parse_string { |
| 641 |
|
| 642 |
my ($attribute, $option) = (@_); |
| 643 |
|
| 644 |
# simple case; return the check_string function. We don't |
| 645 |
# need a destructor for a string so don't return one. |
| 646 |
if (!$option->{sql}) { |
| 647 |
return \&check_string; |
| 648 |
} |
| 649 |
|
| 650 |
=pod |
| 651 |
|
| 652 |
CHAR(N), VARCHAR(N) |
| 653 |
closure checks length of string is less |
| 654 |
than N characters |
| 655 |
|
| 656 |
=cut |
| 657 |
|
| 658 |
if ($option->{sql} =~ m/^\s*(?:var)?char\s*\(\s*(\d+)\s*\)/ix) { |
| 659 |
my $max_length = $1; |
| 660 |
return sub { |
| 661 |
die "string too long for $attribute" |
| 662 |
if (length ${$_[0]} > $max_length); |
| 663 |
}; |
| 664 |
|
| 665 |
=pod |
| 666 |
|
| 667 |
TINYBLOB, BLOB, LONGBLOB |
| 668 |
TINYTEXT, TEXT, LONGTEXT |
| 669 |
checks max. length of string to be 255, |
| 670 |
65535 or 16777215 chars respectively |
| 671 |
|
| 672 |
=cut |
| 673 |
|
| 674 |
} elsif ($option->{sql} =~ m/^\s*(tiny|long|medium)? |
| 675 |
(blob|text)/ix) { |
| 676 |
my $max_length = ($1 ? ($1 eq "tiny"?255:2**24 - 1) |
| 677 |
: 2**16 - 1); |
| 678 |
return sub { |
| 679 |
die "string too long for $attribute" |
| 680 |
if (length ${$_[0]} > $max_length); |
| 681 |
}; |
| 682 |
|
| 683 |
=pod |
| 684 |
|
| 685 |
SET("members", "of", "set") |
| 686 |
checks that the value passed is valid as |
| 687 |
a SQL set type, and that all of the |
| 688 |
passed values are allowed to be a member |
| 689 |
of that set |
| 690 |
|
| 691 |
=cut |
| 692 |
|
| 693 |
} elsif ($option->{sql} =~ |
| 694 |
m/^\s*set\s*\( |
| 695 |
(\"[^\"]+\" (?:\s*,\s*\"[^\"]+\")* \s* ) |
| 696 |
\)\s*$/xi) { |
| 697 |
my $members = $1; |
| 698 |
my ($member, %members); |
| 699 |
while (($member, $members) = |
| 700 |
($members =~ m/^[,\s]*\"([^\"]+)\"(.*)$/)) { |
| 701 |
$members{lc($member)} = 1; |
| 702 |
} |
| 703 |
return sub { |
| 704 |
for my $x (split /,/, ${$_[0]}) { |
| 705 |
croak ("SQL set badly formed or invalid member $x " |
| 706 |
." (SET" . join(",", keys %members). ")") |
| 707 |
if (not exists $members{lc($x)}); |
| 708 |
} |
| 709 |
}; |
| 710 |
|
| 711 |
=pod |
| 712 |
|
| 713 |
ENUM("possible", "values") |
| 714 |
checks that the value passed is one of |
| 715 |
the allowed values. |
| 716 |
|
| 717 |
=cut |
| 718 |
|
| 719 |
} elsif ($option->{sql} =~ |
| 720 |
m/^\s*enum\s*\( |
| 721 |
(\"[^\"]+\" (?:\s*,\s*\"[^\"]+\")* \s* ) |
| 722 |
\)\s*/xi) { |
| 723 |
my $values = $1; |
| 724 |
my ($value, %values); |
| 725 |
while (($value, $values) = |
| 726 |
($values =~ m/^[,\s]*\"([^\"]+)\"(.*)$/)) { |
| 727 |
$values{lc($value)} = 1; |
| 728 |
} |
| 729 |
return sub { |
| 730 |
croak ("invalid enum value ${$_[0]} must be (" |
| 731 |
. join(",", keys %values). ")") |
| 732 |
if (not exists $values{lc(${$_[0]})}); |
| 733 |
} |
| 734 |
|
| 735 |
} else { |
| 736 |
die ("Please build support for your string SQL type in " |
| 737 |
."Class::Tangram (".$option->{sql}.")"); |
| 738 |
} |
| 739 |
} |
| 740 |
|
| 741 |
# Here is where I map Tangram::Type types to functions. |
| 742 |
# Format: |
| 743 |
# type => { |
| 744 |
# check => \&check_X |
| 745 |
# parse => \&parse_type |
| 746 |
# destroy => \&destroy_X |
| 747 |
# } |
| 748 |
# |
| 749 |
my %defaults = |
| 750 |
( |
| 751 |
int => { check => \&check_int }, |
| 752 |
real => { check => \&check_real }, |
| 753 |
string => { parse => \&parse_string }, |
| 754 |
ref => { check => \&check_obj, destroy => \&destroy_ref }, |
| 755 |
array => { check => \&check_array, destroy => \&destroy_array }, |
| 756 |
iarray => { check => \&check_array, destroy => \&destroy_array }, |
| 757 |
flat_array => { check => \&check_flat_array }, |
| 758 |
set => { check => \&check_set, destroy => \&destroy_set }, |
| 759 |
iset => { check => \&check_set, destroy => \&destroy_set }, |
| 760 |
rawdatetime => { check => \&check_rawdatetime }, |
| 761 |
time => { check => \&check_time }, |
| 762 |
timestamp => { check => \&check_timestamp }, |
| 763 |
flat_hash => { check => \&check_flat_hash }, |
| 764 |
hash => { check => \&check_hash, destroy => \&destroy_hash }, |
| 765 |
perl_dump => { check => \&check_nothing } |
| 766 |
); |
| 767 |
|
| 768 |
=item import_schema($class) |
| 769 |
|
| 770 |
Parses a tangram object schema, in "\$${class}::schema" to the |
| 771 |
internal representation used to check types values by set(). Called |
| 772 |
automatically on the first get(), set(), or new() for an object of a |
| 773 |
given class. |
| 774 |
|
| 775 |
This function updates Tangram schema option hashes, with the following |
| 776 |
keys: |
| 777 |
|
| 778 |
check_func - supply/override the check_X function for |
| 779 |
this attribute. |
| 780 |
|
| 781 |
destroy_func - supply/override the destroy_X function for |
| 782 |
this attribute |
| 783 |
|
| 784 |
See the SYNOPSIS section for an example of supplying a check_func in |
| 785 |
an object schema. |
| 786 |
|
| 787 |
=cut |
| 788 |
|
| 789 |
sub import_schema($) { |
| 790 |
my ($class) = (@_); |
| 791 |
|
| 792 |
eval { |
| 793 |
my ($fields, $bases, $abstract); |
| 794 |
{ |
| 795 |
no strict 'refs'; |
| 796 |
$fields = ${"${class}::schema"}->{fields}; |
| 797 |
$bases = ${"${class}::schema"}->{bases}; |
| 798 |
$abstract = ${"${class}::schema"}->{abstract}; |
| 799 |
} |
| 800 |
|
| 801 |
my $check_class = { }; |
| 802 |
my $cleaners_class = { }; |
| 803 |
my $init_defaults_class = { }; |
| 804 |
my $types_class = { }; |
| 805 |
|
| 806 |
# if this is an abstract type, do not allow it to be |
| 807 |
# instantiated |
| 808 |
if ($abstract) { |
| 809 |
$abstract{$class} = 1; |
| 810 |
} |
| 811 |
|
| 812 |
# If there are any base classes, import them first so that the |
| 813 |
# check, cleaners and init_defaults can be inherited |
| 814 |
if (defined $bases) { |
| 815 |
(ref $bases eq "ARRAY") |
| 816 |
or die "bases not an array ref for $class"; |
| 817 |
|
| 818 |
# Note that the order of your bases is significant, that |
| 819 |
# is if you are using multiple iheritance then the later |
| 820 |
# classes override the earlier ones. |
| 821 |
#for my $super ( Class::ISA::super_path($class) ) { |
| 822 |
for my $super ( @$bases ) { |
| 823 |
import_schema $super unless (exists $check{$super}); |
| 824 |
|
| 825 |
# copy each of the per-class configuration hashes to |
| 826 |
# this class as defaults. |
| 827 |
my ($k, $v); |
| 828 |
$types_class->{$k} = $v |
| 829 |
while (($k, $v) = each %{ $types{$super} } ); |
| 830 |
$check_class->{$k} = $v |
| 831 |
while (($k, $v) = each %{ $check{$super} } ); |
| 832 |
$cleaners_class->{$k} = $v |
| 833 |
while (($k, $v) = each %{ $cleaners{$super} } ); |
| 834 |
$init_defaults_class->{$k} = $v |
| 835 |
while (($k, $v) = each %{ $init_defaults{$super} } ); |
| 836 |
} |
| 837 |
} |
| 838 |
|
| 839 |
# iterate over each of the *types* of fields (string, int, ref, etc.) |
| 840 |
while (my ($type, $v) = each %$fields) { |
| 841 |
if (ref $v eq "ARRAY") { |
| 842 |
$v = { map { $_, undef } @$v }; |
| 843 |
} |
| 844 |
my $def = $defaults{$type}; |
| 845 |
|
| 846 |
# iterate each of the *attributes* of a particular type |
| 847 |
while (my ($attribute, $option) = each %$v) { |
| 848 |
$types_class->{$attribute} = $type; |
| 849 |
|
| 850 |
# ----- check_X functions ---- |
| 851 |
if (ref $option eq "HASH" and $option->{check_func}) { |
| 852 |
# user-supplied check_X function |
| 853 |
$check_class->{$attribute} = |
| 854 |
$option->{check_func}; |
| 855 |
|
| 856 |
} else { |
| 857 |
if (not defined $def) { |
| 858 |
die "No default check function for type $type"; |
| 859 |
} |
| 860 |
|
| 861 |
# check for a type-specific option hash parser |
| 862 |
if ($def->{parse}) { |
| 863 |
my ($check, $destroy) = |
| 864 |
$def->{parse}->($attribute, $option); |
| 865 |
|
| 866 |
$check_class->{$attribute} = $check; |
| 867 |
$cleaners_class->{$attribute} = $destroy |
| 868 |
if (defined $destroy); |
| 869 |
|
| 870 |
} else { |
| 871 |
# use the default for this type |
| 872 |
$check_class->{$attribute} = $def->{check}; |
| 873 |
} |
| 874 |
} |
| 875 |
|
| 876 |
# ----- destroy_X functions |
| 877 |
if (ref $option eq "HASH" and $option->{destroy_func}) { |
| 878 |
# user-supplied destroy_X function |
| 879 |
$cleaners_class->{$attribute} = |
| 880 |
$option->{destroy_func}; |
| 881 |
} else { |
| 882 |
if ($def->{destroy}) { |
| 883 |
# use the default for this type |
| 884 |
$cleaners_class->{$attribute} = |
| 885 |
$def->{destroy}; |
| 886 |
} |
| 887 |
} |
| 888 |
|
| 889 |
# ----- init_default functions |
| 890 |
# create empty Set::Object containers as necessary |
| 891 |
if ($type =~ m/^i?set$/) { |
| 892 |
$init_defaults_class->{$attribute} = |
| 893 |
sub { Set::Object->new() }; |
| 894 |
} |
| 895 |
if (ref $option eq "HASH" and $option->{init_default}) { |
| 896 |
$init_defaults_class->{$attribute} = |
| 897 |
$option->{init_default}; |
| 898 |
} |
| 899 |
} |
| 900 |
} |
| 901 |
$types{$class} = $types_class; |
| 902 |
$check{$class} = $check_class; |
| 903 |
$cleaners{$class} = $cleaners_class; |
| 904 |
$init_defaults{$class} = $init_defaults_class; |
| 905 |
}; |
| 906 |
|
| 907 |
$@ && die "$@ while trying to import schema for $class"; |
| 908 |
} |
| 909 |
|
| 910 |
|
| 911 |
=item $instance->quickdump |
| 912 |
|
| 913 |
Quickly show the blessed hash of an object, without descending into |
| 914 |
it. Primarily useful when you have a large interconnected graph of |
| 915 |
objects so don't want to use the B<x> command within the debugger. |
| 916 |
It also doesn't have the side effect of auto-vivifying members. |
| 917 |
|
| 918 |
This function returns a string, suitable for print()ing. It does not |
| 919 |
currently escape unprintable characters. |
| 920 |
|
| 921 |
=cut |
| 922 |
|
| 923 |
sub quickdump($) { |
| 924 |
my ($self) = (@_); |
| 925 |
|
| 926 |
my $r = "REF ". (ref $self). "\n"; |
| 927 |
for my $k (sort keys %$self) { |
| 928 |
$r .= (" $k => " |
| 929 |
. ( |
| 930 |
tied $self->{$k} |
| 931 |
|| ( ref $self->{$k} |
| 932 |
? $self->{$k} |
| 933 |
: "'".$self->{$k}."'" ) |
| 934 |
) |
| 935 |
. "\n"); |
| 936 |
} |
| 937 |
return $r; |
| 938 |
} |
| 939 |
|
| 940 |
|
| 941 |
=item $instance->DESTROY |
| 942 |
|
| 943 |
This function ensures that all of your attributes have their |
| 944 |
destructors called. It calls the destroy_X function for attributes |
| 945 |
that have it defined, if that attribute exists in the instance that we |
| 946 |
are destroying. It calls the destroy_X functions as destroy_X($self, |
| 947 |
$k) |
| 948 |
|
| 949 |
=cut |
| 950 |
|
| 951 |
sub DESTROY($) { |
| 952 |
my ($self) = (@_); |
| 953 |
|
| 954 |
my $class = ref $self; |
| 955 |
|
| 956 |
# if no cleaners are known for this class, it hasn't been imported |
| 957 |
# yet. Don't call import_schema, that would be a bad idea in a |
| 958 |
# destructor. |
| 959 |
exists $cleaners{$class} or return; |
| 960 |
|
| 961 |
# for every attribute that is defined, and has a cleaner function, |
| 962 |
# call the cleaner function. |
| 963 |
for my $k (keys %$self) { |
| 964 |
if (defined $cleaners{$class}->{$k} and exists $self->{$k}) { |
| 965 |
$cleaners{$class}->{$k}->($self, $k); |
| 966 |
} |
| 967 |
} |
| 968 |
$self->{_DESTROYED} = 1; |
| 969 |
} |
| 970 |
|
| 971 |
=item $instance->clear_refs |
| 972 |
|
| 973 |
This clears all references from this object, ie exactly what DESTROY |
| 974 |
normally does, but calling an object's destructor method directly is |
| 975 |
bad form. Also, this function has no qualms with loading the class' |
| 976 |
schema with import_schema() as needed. |
| 977 |
|
| 978 |
This is useful for breaking circular references, if you know you are |
| 979 |
no longer going to be using an object then you can call this method, |
| 980 |
which in many cases will end up cleaning up most of the objects you |
| 981 |
want to get rid of. |
| 982 |
|
| 983 |
However, it still won't do anything about Tangram's internal reference |
| 984 |
to the object, which must still be explicitly unlinked with the |
| 985 |
Tangram::Storage->unload method. |
| 986 |
|
| 987 |
=cut |
| 988 |
|
| 989 |
sub clear_refs($) { |
| 990 |
my ($self) = (@_); |
| 991 |
|
| 992 |
my $class = ref $self; |
| 993 |
|
| 994 |
exists $cleaners{$class} or import_schema($class); |
| 995 |
|
| 996 |
# break all ref's, sets, arrays |
| 997 |
for my $k (keys %$self) { |
| 998 |
if (defined $cleaners{$class}->{$k} and exists $self->{$k}) { |
| 999 |
$cleaners{$class}->{$k}->($self, $k); |
| 1000 |
} |
| 1001 |
} |
| 1002 |
$self->{_NOREFS} = 1; |
| 1003 |
} |
| 1004 |
|
| 1005 |
|
| 1006 |
=back |
| 1007 |
|
| 1008 |
=head1 SEE ALSO |
| 1009 |
|
| 1010 |
L<Tangram::Schema> |
| 1011 |
|
| 1012 |
B<A guided tour of Tangram, by Sound Object Logic.> |
| 1013 |
|
| 1014 |
http://www.soundobjectlogic.com/tangram/guided_tour/fs.html |
| 1015 |
|
| 1016 |
=head1 BUGS/TODO |
| 1017 |
|
| 1018 |
More datetime types. I originally avoided the DMDateTime type because |
| 1019 |
Date::Manip is self-admittedly the most bloated module on CPAN, and I |
| 1020 |
didn't want to be seen encouraging it. Then I found out about |
| 1021 |
autosplit :-}. |
| 1022 |
|
| 1023 |
More AUTOLOAD methods, in particular for container types such as |
| 1024 |
array, hash, etc. |
| 1025 |
|
| 1026 |
This documentation should be easy enough for a fool to understand. |
| 1027 |
|
| 1028 |
There should be more functions for breaking loops; in particular, a |
| 1029 |
standard function called drop_refs($obj), which replaces references to |
| 1030 |
$obj with the appropriate Tangram::RefOnDemand so that an object can |
| 1031 |
be unloaded via Tangram::Storage->unload() and actually have a hope of |
| 1032 |
being reclaimed. Another function that would be handy would be a |
| 1033 |
deep "mark" operation for mark & sweep garbage collection. |
| 1034 |
|
| 1035 |
=head1 AUTHOR |
| 1036 |
|
| 1037 |
Sam Vilain, <sam@vilain.net> |
| 1038 |
|
| 1039 |
=cut |
| 1040 |
|
| 1041 |
69; |
Parent Directory
| 
Revision Log
