Commit | Line | Data |
25f64a11 |
1 | # |
2 | # Copyright (c) 1995-2000, Raphael Manfredi |
3 | # |
4 | # You may redistribute only under the same terms as Perl 5, as specified |
5 | # in the README file that comes with the distribution. |
6 | # |
7a6a85bf |
7 | |
8 | require DynaLoader; |
9 | require Exporter; |
10 | package Storable; @ISA = qw(Exporter DynaLoader); |
11 | |
12 | @EXPORT = qw(store retrieve); |
13 | @EXPORT_OK = qw( |
9e21b3d0 |
14 | nstore store_fd nstore_fd fd_retrieve |
7a6a85bf |
15 | freeze nfreeze thaw |
16 | dclone |
9e21b3d0 |
17 | retrieve_fd |
dd19458b |
18 | lock_store lock_nstore lock_retrieve |
7a6a85bf |
19 | ); |
20 | |
21 | use AutoLoader; |
01d7b99e |
22 | use vars qw($canonical $forgive_me $VERSION); |
7a6a85bf |
23 | |
3f575d8d |
24 | $VERSION = '2.12'; |
7a6a85bf |
25 | *AUTOLOAD = \&AutoLoader::AUTOLOAD; # Grrr... |
26 | |
27 | # |
28 | # Use of Log::Agent is optional |
29 | # |
30 | |
31 | eval "use Log::Agent"; |
32 | |
530b72ba |
33 | require Carp; |
7a6a85bf |
34 | |
dd19458b |
35 | # |
36 | # They might miss :flock in Fcntl |
37 | # |
38 | |
39 | BEGIN { |
596596d5 |
40 | if (eval { require Fcntl; 1 } && exists $Fcntl::EXPORT_TAGS{'flock'}) { |
dd19458b |
41 | Fcntl->import(':flock'); |
42 | } else { |
43 | eval q{ |
44 | sub LOCK_SH () {1} |
45 | sub LOCK_EX () {2} |
46 | }; |
47 | } |
48 | } |
49 | |
a8b7ef86 |
50 | sub CLONE { |
51 | # clone context under threads |
52 | Storable::init_perinterp(); |
53 | } |
54 | |
b8778c7c |
55 | # Can't Autoload cleanly as this clashes 8.3 with &retrieve |
9e21b3d0 |
56 | sub retrieve_fd { &fd_retrieve } # Backward compatibility |
cb3d9de5 |
57 | |
530b72ba |
58 | # By default restricted hashes are downgraded on earlier perls. |
59 | |
60 | $Storable::downgrade_restricted = 1; |
e8189732 |
61 | $Storable::accept_future_minor = 1; |
b8778c7c |
62 | bootstrap Storable; |
63 | 1; |
64 | __END__ |
530b72ba |
65 | # |
66 | # Use of Log::Agent is optional. If it hasn't imported these subs then |
67 | # Autoloader will kindly supply our fallback implementation. |
68 | # |
69 | |
70 | sub logcroak { |
71 | Carp::croak(@_); |
72 | } |
73 | |
74 | sub logcarp { |
75 | Carp::carp(@_); |
76 | } |
b8778c7c |
77 | |
862382c7 |
78 | # |
79 | # Determine whether locking is possible, but only when needed. |
80 | # |
81 | |
530b72ba |
82 | sub CAN_FLOCK; my $CAN_FLOCK; sub CAN_FLOCK { |
862382c7 |
83 | return $CAN_FLOCK if defined $CAN_FLOCK; |
84 | require Config; import Config; |
85 | return $CAN_FLOCK = |
86 | $Config{'d_flock'} || |
87 | $Config{'d_fcntl_can_lock'} || |
88 | $Config{'d_lockf'}; |
89 | } |
90 | |
0a0da639 |
91 | sub show_file_magic { |
92 | print <<EOM; |
93 | # |
94 | # To recognize the data files of the Perl module Storable, |
95 | # the following lines need to be added to the local magic(5) file, |
96 | # usually either /usr/share/misc/magic or /etc/magic. |
0a0da639 |
97 | # |
98 | 0 string perl-store perl Storable(v0.6) data |
8b793558 |
99 | >4 byte >0 (net-order %d) |
100 | >>4 byte &01 (network-ordered) |
101 | >>4 byte =3 (major 1) |
102 | >>4 byte =2 (major 1) |
103 | |
0a0da639 |
104 | 0 string pst0 perl Storable(v0.7) data |
8b793558 |
105 | >4 byte >0 |
106 | >>4 byte &01 (network-ordered) |
107 | >>4 byte =5 (major 2) |
108 | >>4 byte =4 (major 2) |
109 | >>5 byte >0 (minor %d) |
0a0da639 |
110 | EOM |
111 | } |
112 | |
b8778c7c |
113 | sub read_magic { |
114 | my $header = shift; |
115 | return unless defined $header and length $header > 11; |
116 | my $result; |
117 | if ($header =~ s/^perl-store//) { |
118 | die "Can't deal with version 0 headers"; |
119 | } elsif ($header =~ s/^pst0//) { |
120 | $result->{file} = 1; |
121 | } |
122 | # Assume it's a string. |
123 | my ($major, $minor, $bytelen) = unpack "C3", $header; |
124 | |
125 | my $net_order = $major & 1; |
126 | $major >>= 1; |
127 | @$result{qw(major minor netorder)} = ($major, $minor, $net_order); |
128 | |
129 | return $result if $net_order; |
130 | |
131 | # I assume that it is rare to find v1 files, so this is an intentionally |
132 | # inefficient way of doing it, to make the rest of the code constant. |
133 | if ($major < 2) { |
134 | delete $result->{minor}; |
135 | $header = '.' . $header; |
136 | $bytelen = $minor; |
137 | } |
138 | |
139 | @$result{qw(byteorder intsize longsize ptrsize)} = |
140 | unpack "x3 A$bytelen C3", $header; |
141 | |
142 | if ($major >= 2 and $minor >= 2) { |
143 | $result->{nvsize} = unpack "x6 x$bytelen C", $header; |
144 | } |
145 | $result; |
146 | } |
7a6a85bf |
147 | |
148 | # |
149 | # store |
150 | # |
151 | # Store target object hierarchy, identified by a reference to its root. |
152 | # The stored object tree may later be retrieved to memory via retrieve. |
153 | # Returns undef if an I/O error occurred, in which case the file is |
154 | # removed. |
155 | # |
156 | sub store { |
dd19458b |
157 | return _store(\&pstore, @_, 0); |
7a6a85bf |
158 | } |
159 | |
160 | # |
161 | # nstore |
162 | # |
163 | # Same as store, but in network order. |
164 | # |
165 | sub nstore { |
dd19458b |
166 | return _store(\&net_pstore, @_, 0); |
167 | } |
168 | |
169 | # |
170 | # lock_store |
171 | # |
172 | # Same as store, but flock the file first (advisory locking). |
173 | # |
174 | sub lock_store { |
175 | return _store(\&pstore, @_, 1); |
176 | } |
177 | |
178 | # |
179 | # lock_nstore |
180 | # |
181 | # Same as nstore, but flock the file first (advisory locking). |
182 | # |
183 | sub lock_nstore { |
184 | return _store(\&net_pstore, @_, 1); |
7a6a85bf |
185 | } |
186 | |
187 | # Internal store to file routine |
188 | sub _store { |
189 | my $xsptr = shift; |
190 | my $self = shift; |
dd19458b |
191 | my ($file, $use_locking) = @_; |
7a6a85bf |
192 | logcroak "not a reference" unless ref($self); |
b12202d0 |
193 | logcroak "wrong argument number" unless @_ == 2; # No @foo in arglist |
7a6a85bf |
194 | local *FILE; |
dd19458b |
195 | if ($use_locking) { |
6e0ac6f5 |
196 | open(FILE, ">>$file") || logcroak "can't write into $file: $!"; |
862382c7 |
197 | unless (&CAN_FLOCK) { |
b29b780f |
198 | logcarp "Storable::lock_store: fcntl/flock emulation broken on $^O"; |
199 | return undef; |
f567092b |
200 | } |
dd19458b |
201 | flock(FILE, LOCK_EX) || |
202 | logcroak "can't get exclusive lock on $file: $!"; |
203 | truncate FILE, 0; |
204 | # Unlocking will happen when FILE is closed |
6e0ac6f5 |
205 | } else { |
206 | open(FILE, ">$file") || logcroak "can't create $file: $!"; |
dd19458b |
207 | } |
6e0ac6f5 |
208 | binmode FILE; # Archaic systems... |
7a6a85bf |
209 | my $da = $@; # Don't mess if called from exception handler |
210 | my $ret; |
211 | # Call C routine nstore or pstore, depending on network order |
212 | eval { $ret = &$xsptr(*FILE, $self) }; |
213 | close(FILE) or $ret = undef; |
214 | unlink($file) or warn "Can't unlink $file: $!\n" if $@ || !defined $ret; |
215 | logcroak $@ if $@ =~ s/\.?\n$/,/; |
216 | $@ = $da; |
217 | return $ret ? $ret : undef; |
218 | } |
219 | |
220 | # |
221 | # store_fd |
222 | # |
223 | # Same as store, but perform on an already opened file descriptor instead. |
224 | # Returns undef if an I/O error occurred. |
225 | # |
226 | sub store_fd { |
227 | return _store_fd(\&pstore, @_); |
228 | } |
229 | |
230 | # |
231 | # nstore_fd |
232 | # |
233 | # Same as store_fd, but in network order. |
234 | # |
235 | sub nstore_fd { |
236 | my ($self, $file) = @_; |
237 | return _store_fd(\&net_pstore, @_); |
238 | } |
239 | |
240 | # Internal store routine on opened file descriptor |
241 | sub _store_fd { |
242 | my $xsptr = shift; |
243 | my $self = shift; |
244 | my ($file) = @_; |
245 | logcroak "not a reference" unless ref($self); |
246 | logcroak "too many arguments" unless @_ == 1; # No @foo in arglist |
247 | my $fd = fileno($file); |
248 | logcroak "not a valid file descriptor" unless defined $fd; |
249 | my $da = $@; # Don't mess if called from exception handler |
250 | my $ret; |
251 | # Call C routine nstore or pstore, depending on network order |
252 | eval { $ret = &$xsptr($file, $self) }; |
253 | logcroak $@ if $@ =~ s/\.?\n$/,/; |
596596d5 |
254 | local $\; print $file ''; # Autoflush the file if wanted |
7a6a85bf |
255 | $@ = $da; |
256 | return $ret ? $ret : undef; |
257 | } |
258 | |
259 | # |
260 | # freeze |
261 | # |
262 | # Store oject and its hierarchy in memory and return a scalar |
263 | # containing the result. |
264 | # |
265 | sub freeze { |
266 | _freeze(\&mstore, @_); |
267 | } |
268 | |
269 | # |
270 | # nfreeze |
271 | # |
272 | # Same as freeze but in network order. |
273 | # |
274 | sub nfreeze { |
275 | _freeze(\&net_mstore, @_); |
276 | } |
277 | |
278 | # Internal freeze routine |
279 | sub _freeze { |
280 | my $xsptr = shift; |
281 | my $self = shift; |
282 | logcroak "not a reference" unless ref($self); |
283 | logcroak "too many arguments" unless @_ == 0; # No @foo in arglist |
284 | my $da = $@; # Don't mess if called from exception handler |
285 | my $ret; |
286 | # Call C routine mstore or net_mstore, depending on network order |
287 | eval { $ret = &$xsptr($self) }; |
288 | logcroak $@ if $@ =~ s/\.?\n$/,/; |
289 | $@ = $da; |
290 | return $ret ? $ret : undef; |
291 | } |
292 | |
293 | # |
294 | # retrieve |
295 | # |
296 | # Retrieve object hierarchy from disk, returning a reference to the root |
297 | # object of that tree. |
298 | # |
299 | sub retrieve { |
dd19458b |
300 | _retrieve($_[0], 0); |
301 | } |
302 | |
303 | # |
304 | # lock_retrieve |
305 | # |
306 | # Same as retrieve, but with advisory locking. |
307 | # |
308 | sub lock_retrieve { |
309 | _retrieve($_[0], 1); |
310 | } |
311 | |
312 | # Internal retrieve routine |
313 | sub _retrieve { |
314 | my ($file, $use_locking) = @_; |
7a6a85bf |
315 | local *FILE; |
dd19458b |
316 | open(FILE, $file) || logcroak "can't open $file: $!"; |
7a6a85bf |
317 | binmode FILE; # Archaic systems... |
318 | my $self; |
319 | my $da = $@; # Could be from exception handler |
dd19458b |
320 | if ($use_locking) { |
862382c7 |
321 | unless (&CAN_FLOCK) { |
8be2b38b |
322 | logcarp "Storable::lock_store: fcntl/flock emulation broken on $^O"; |
b29b780f |
323 | return undef; |
324 | } |
8be2b38b |
325 | flock(FILE, LOCK_SH) || logcroak "can't get shared lock on $file: $!"; |
dd19458b |
326 | # Unlocking will happen when FILE is closed |
327 | } |
7a6a85bf |
328 | eval { $self = pretrieve(*FILE) }; # Call C routine |
329 | close(FILE); |
330 | logcroak $@ if $@ =~ s/\.?\n$/,/; |
331 | $@ = $da; |
332 | return $self; |
333 | } |
334 | |
335 | # |
9e21b3d0 |
336 | # fd_retrieve |
7a6a85bf |
337 | # |
338 | # Same as retrieve, but perform from an already opened file descriptor instead. |
339 | # |
9e21b3d0 |
340 | sub fd_retrieve { |
7a6a85bf |
341 | my ($file) = @_; |
342 | my $fd = fileno($file); |
343 | logcroak "not a valid file descriptor" unless defined $fd; |
344 | my $self; |
345 | my $da = $@; # Could be from exception handler |
346 | eval { $self = pretrieve($file) }; # Call C routine |
347 | logcroak $@ if $@ =~ s/\.?\n$/,/; |
348 | $@ = $da; |
349 | return $self; |
350 | } |
351 | |
352 | # |
353 | # thaw |
354 | # |
355 | # Recreate objects in memory from an existing frozen image created |
356 | # by freeze. If the frozen image passed is undef, return undef. |
357 | # |
358 | sub thaw { |
359 | my ($frozen) = @_; |
360 | return undef unless defined $frozen; |
361 | my $self; |
362 | my $da = $@; # Could be from exception handler |
363 | eval { $self = mretrieve($frozen) }; # Call C routine |
364 | logcroak $@ if $@ =~ s/\.?\n$/,/; |
365 | $@ = $da; |
366 | return $self; |
367 | } |
368 | |
a2307be4 |
369 | 1; |
370 | __END__ |
371 | |
7a6a85bf |
372 | =head1 NAME |
373 | |
f062ea6c |
374 | Storable - persistence for Perl data structures |
7a6a85bf |
375 | |
376 | =head1 SYNOPSIS |
377 | |
378 | use Storable; |
379 | store \%table, 'file'; |
380 | $hashref = retrieve('file'); |
381 | |
382 | use Storable qw(nstore store_fd nstore_fd freeze thaw dclone); |
383 | |
384 | # Network order |
385 | nstore \%table, 'file'; |
386 | $hashref = retrieve('file'); # There is NO nretrieve() |
387 | |
388 | # Storing to and retrieving from an already opened file |
389 | store_fd \@array, \*STDOUT; |
390 | nstore_fd \%table, \*STDOUT; |
9e21b3d0 |
391 | $aryref = fd_retrieve(\*SOCKET); |
392 | $hashref = fd_retrieve(\*SOCKET); |
7a6a85bf |
393 | |
394 | # Serializing to memory |
395 | $serialized = freeze \%table; |
396 | %table_clone = %{ thaw($serialized) }; |
397 | |
398 | # Deep (recursive) cloning |
399 | $cloneref = dclone($ref); |
400 | |
dd19458b |
401 | # Advisory locking |
402 | use Storable qw(lock_store lock_nstore lock_retrieve) |
403 | lock_store \%table, 'file'; |
404 | lock_nstore \%table, 'file'; |
405 | $hashref = lock_retrieve('file'); |
406 | |
7a6a85bf |
407 | =head1 DESCRIPTION |
408 | |
f062ea6c |
409 | The Storable package brings persistence to your Perl data structures |
7a6a85bf |
410 | containing SCALAR, ARRAY, HASH or REF objects, i.e. anything that can be |
c261f00e |
411 | conveniently stored to disk and retrieved at a later time. |
7a6a85bf |
412 | |
413 | It can be used in the regular procedural way by calling C<store> with |
414 | a reference to the object to be stored, along with the file name where |
415 | the image should be written. |
775ecd75 |
416 | |
7a6a85bf |
417 | The routine returns C<undef> for I/O problems or other internal error, |
418 | a true value otherwise. Serious errors are propagated as a C<die> exception. |
419 | |
f062ea6c |
420 | To retrieve data stored to disk, use C<retrieve> with a file name. |
421 | The objects stored into that file are recreated into memory for you, |
422 | and a I<reference> to the root object is returned. In case an I/O error |
7a6a85bf |
423 | occurs while reading, C<undef> is returned instead. Other serious |
424 | errors are propagated via C<die>. |
425 | |
426 | Since storage is performed recursively, you might want to stuff references |
427 | to objects that share a lot of common data into a single array or hash |
428 | table, and then store that object. That way, when you retrieve back the |
429 | whole thing, the objects will continue to share what they originally shared. |
430 | |
431 | At the cost of a slight header overhead, you may store to an already |
432 | opened file descriptor using the C<store_fd> routine, and retrieve |
9e21b3d0 |
433 | from a file via C<fd_retrieve>. Those names aren't imported by default, |
c261f00e |
434 | so you will have to do that explicitly if you need those routines. |
7a6a85bf |
435 | The file descriptor you supply must be already opened, for read |
436 | if you're going to retrieve and for write if you wish to store. |
437 | |
438 | store_fd(\%table, *STDOUT) || die "can't store to stdout\n"; |
9e21b3d0 |
439 | $hashref = fd_retrieve(*STDIN); |
7a6a85bf |
440 | |
441 | You can also store data in network order to allow easy sharing across |
442 | multiple platforms, or when storing on a socket known to be remotely |
443 | connected. The routines to call have an initial C<n> prefix for I<network>, |
444 | as in C<nstore> and C<nstore_fd>. At retrieval time, your data will be |
445 | correctly restored so you don't have to know whether you're restoring |
dd19458b |
446 | from native or network ordered data. Double values are stored stringified |
447 | to ensure portability as well, at the slight risk of loosing some precision |
448 | in the last decimals. |
7a6a85bf |
449 | |
9e21b3d0 |
450 | When using C<fd_retrieve>, objects are retrieved in sequence, one |
7a6a85bf |
451 | object (i.e. one recursive tree) per associated C<store_fd>. |
452 | |
453 | If you're more from the object-oriented camp, you can inherit from |
454 | Storable and directly store your objects by invoking C<store> as |
455 | a method. The fact that the root of the to-be-stored tree is a |
456 | blessed reference (i.e. an object) is special-cased so that the |
457 | retrieve does not provide a reference to that object but rather the |
458 | blessed object reference itself. (Otherwise, you'd get a reference |
459 | to that blessed object). |
460 | |
461 | =head1 MEMORY STORE |
462 | |
463 | The Storable engine can also store data into a Perl scalar instead, to |
464 | later retrieve them. This is mainly used to freeze a complex structure in |
465 | some safe compact memory place (where it can possibly be sent to another |
466 | process via some IPC, since freezing the structure also serializes it in |
467 | effect). Later on, and maybe somewhere else, you can thaw the Perl scalar |
468 | out and recreate the original complex structure in memory. |
469 | |
470 | Surprisingly, the routines to be called are named C<freeze> and C<thaw>. |
471 | If you wish to send out the frozen scalar to another machine, use |
472 | C<nfreeze> instead to get a portable image. |
473 | |
474 | Note that freezing an object structure and immediately thawing it |
475 | actually achieves a deep cloning of that structure: |
476 | |
477 | dclone(.) = thaw(freeze(.)) |
478 | |
479 | Storable provides you with a C<dclone> interface which does not create |
480 | that intermediary scalar but instead freezes the structure in some |
c261f00e |
481 | internal memory space and then immediately thaws it out. |
7a6a85bf |
482 | |
dd19458b |
483 | =head1 ADVISORY LOCKING |
484 | |
f062ea6c |
485 | The C<lock_store> and C<lock_nstore> routine are equivalent to |
486 | C<store> and C<nstore>, except that they get an exclusive lock on |
487 | the file before writing. Likewise, C<lock_retrieve> does the same |
488 | as C<retrieve>, but also gets a shared lock on the file before reading. |
dd19458b |
489 | |
f062ea6c |
490 | As with any advisory locking scheme, the protection only works if you |
491 | systematically use C<lock_store> and C<lock_retrieve>. If one side of |
492 | your application uses C<store> whilst the other uses C<lock_retrieve>, |
dd19458b |
493 | you will get no protection at all. |
494 | |
f062ea6c |
495 | The internal advisory locking is implemented using Perl's flock() |
496 | routine. If your system does not support any form of flock(), or if |
497 | you share your files across NFS, you might wish to use other forms |
498 | of locking by using modules such as LockFile::Simple which lock a |
499 | file using a filesystem entry, instead of locking the file descriptor. |
dd19458b |
500 | |
7a6a85bf |
501 | =head1 SPEED |
502 | |
503 | The heart of Storable is written in C for decent speed. Extra low-level |
4d3295e3 |
504 | optimizations have been made when manipulating perl internals, to |
505 | sacrifice encapsulation for the benefit of greater speed. |
7a6a85bf |
506 | |
507 | =head1 CANONICAL REPRESENTATION |
508 | |
f062ea6c |
509 | Normally, Storable stores elements of hashes in the order they are |
7a6a85bf |
510 | stored internally by Perl, i.e. pseudo-randomly. If you set |
511 | C<$Storable::canonical> to some C<TRUE> value, Storable will store |
512 | hashes with the elements sorted by their key. This allows you to |
513 | compare data structures by comparing their frozen representations (or |
514 | even the compressed frozen representations), which can be useful for |
515 | creating lookup tables for complicated queries. |
516 | |
f062ea6c |
517 | Canonical order does not imply network order; those are two orthogonal |
7a6a85bf |
518 | settings. |
519 | |
d2b96869 |
520 | =head1 CODE REFERENCES |
521 | |
522 | Since Storable version 2.05, CODE references may be serialized with |
523 | the help of L<B::Deparse>. To enable this feature, set |
524 | C<$Storable::Deparse> to a true value. To enable deserializazion, |
525 | C<$Storable::Eval> should be set to a true value. Be aware that |
526 | deserialization is done through C<eval>, which is dangerous if the |
527 | Storable file contains malicious data. You can set C<$Storable::Eval> |
528 | to a subroutine reference which would be used instead of C<eval>. See |
529 | below for an example using a L<Safe> compartment for deserialization |
530 | of CODE references. |
531 | |
197b90bc |
532 | If C<$Storable::Deparse> and/or C<$Storable::Eval> are set to false |
533 | values, then the value of C<$Storable::forgive_me> (see below) is |
534 | respected while serializing and deserializing. |
535 | |
c261f00e |
536 | =head1 FORWARD COMPATIBILITY |
537 | |
538 | This release of Storable can be used on a newer version of Perl to |
f062ea6c |
539 | serialize data which is not supported by earlier Perls. By default, |
c261f00e |
540 | Storable will attempt to do the right thing, by C<croak()>ing if it |
775ecd75 |
541 | encounters data that it cannot deserialize. However, the defaults |
f062ea6c |
542 | can be changed as follows: |
c261f00e |
543 | |
544 | =over 4 |
545 | |
546 | =item utf8 data |
547 | |
548 | Perl 5.6 added support for Unicode characters with code points > 255, |
549 | and Perl 5.8 has full support for Unicode characters in hash keys. |
550 | Perl internally encodes strings with these characters using utf8, and |
551 | Storable serializes them as utf8. By default, if an older version of |
552 | Perl encounters a utf8 value it cannot represent, it will C<croak()>. |
553 | To change this behaviour so that Storable deserializes utf8 encoded |
554 | values as the string of bytes (effectively dropping the I<is_utf8> flag) |
555 | set C<$Storable::drop_utf8> to some C<TRUE> value. This is a form of |
556 | data loss, because with C<$drop_utf8> true, it becomes impossible to tell |
557 | whether the original data was the Unicode string, or a series of bytes |
558 | that happen to be valid utf8. |
559 | |
560 | =item restricted hashes |
561 | |
f062ea6c |
562 | Perl 5.8 adds support for restricted hashes, which have keys |
563 | restricted to a given set, and can have values locked to be read only. |
564 | By default, when Storable encounters a restricted hash on a perl |
565 | that doesn't support them, it will deserialize it as a normal hash, |
566 | silently discarding any placeholder keys and leaving the keys and |
567 | all values unlocked. To make Storable C<croak()> instead, set |
568 | C<$Storable::downgrade_restricted> to a C<FALSE> value. To restore |
569 | the default set it back to some C<TRUE> value. |
c261f00e |
570 | |
e8189732 |
571 | =item files from future versions of Storable |
572 | |
573 | Earlier versions of Storable would immediately croak if they encountered |
574 | a file with a higher internal version number than the reading Storable |
575 | knew about. Internal version numbers are increased each time new data |
576 | types (such as restricted hashes) are added to the vocabulary of the file |
577 | format. This meant that a newer Storable module had no way of writing a |
f062ea6c |
578 | file readable by an older Storable, even if the writer didn't store newer |
e8189732 |
579 | data types. |
580 | |
581 | This version of Storable will defer croaking until it encounters a data |
582 | type in the file that it does not recognize. This means that it will |
583 | continue to read files generated by newer Storable modules which are careful |
584 | in what they write out, making it easier to upgrade Storable modules in a |
585 | mixed environment. |
586 | |
587 | The old behaviour of immediate croaking can be re-instated by setting |
f062ea6c |
588 | C<$Storable::accept_future_minor> to some C<FALSE> value. |
e8189732 |
589 | |
c261f00e |
590 | =back |
591 | |
f062ea6c |
592 | All these variables have no effect on a newer Perl which supports the |
c261f00e |
593 | relevant feature. |
594 | |
7a6a85bf |
595 | =head1 ERROR REPORTING |
596 | |
597 | Storable uses the "exception" paradigm, in that it does not try to workaround |
598 | failures: if something bad happens, an exception is generated from the |
599 | caller's perspective (see L<Carp> and C<croak()>). Use eval {} to trap |
600 | those exceptions. |
601 | |
602 | When Storable croaks, it tries to report the error via the C<logcroak()> |
603 | routine from the C<Log::Agent> package, if it is available. |
604 | |
212e9bde |
605 | Normal errors are reported by having store() or retrieve() return C<undef>. |
606 | Such errors are usually I/O errors (or truncated stream errors at retrieval). |
607 | |
7a6a85bf |
608 | =head1 WIZARDS ONLY |
609 | |
610 | =head2 Hooks |
611 | |
612 | Any class may define hooks that will be called during the serialization |
613 | and deserialization process on objects that are instances of that class. |
614 | Those hooks can redefine the way serialization is performed (and therefore, |
c261f00e |
615 | how the symmetrical deserialization should be conducted). |
7a6a85bf |
616 | |
617 | Since we said earlier: |
618 | |
619 | dclone(.) = thaw(freeze(.)) |
620 | |
621 | everything we say about hooks should also hold for deep cloning. However, |
622 | hooks get to know whether the operation is a mere serialization, or a cloning. |
623 | |
624 | Therefore, when serializing hooks are involved, |
625 | |
626 | dclone(.) <> thaw(freeze(.)) |
627 | |
628 | Well, you could keep them in sync, but there's no guarantee it will always |
629 | hold on classes somebody else wrote. Besides, there is little to gain in |
f062ea6c |
630 | doing so: a serializing hook could keep only one attribute of an object, |
7a6a85bf |
631 | which is probably not what should happen during a deep cloning of that |
632 | same object. |
633 | |
634 | Here is the hooking interface: |
635 | |
bbc7dcd2 |
636 | =over 4 |
7a6a85bf |
637 | |
638 | =item C<STORABLE_freeze> I<obj>, I<cloning> |
639 | |
640 | The serializing hook, called on the object during serialization. It can be |
641 | inherited, or defined in the class itself, like any other method. |
642 | |
643 | Arguments: I<obj> is the object to serialize, I<cloning> is a flag indicating |
644 | whether we're in a dclone() or a regular serialization via store() or freeze(). |
645 | |
646 | Returned value: A LIST C<($serialized, $ref1, $ref2, ...)> where $serialized |
647 | is the serialized form to be used, and the optional $ref1, $ref2, etc... are |
648 | extra references that you wish to let the Storable engine serialize. |
649 | |
650 | At deserialization time, you will be given back the same LIST, but all the |
651 | extra references will be pointing into the deserialized structure. |
652 | |
653 | The B<first time> the hook is hit in a serialization flow, you may have it |
654 | return an empty list. That will signal the Storable engine to further |
655 | discard that hook for this class and to therefore revert to the default |
656 | serialization of the underlying Perl data. The hook will again be normally |
657 | processed in the next serialization. |
658 | |
659 | Unless you know better, serializing hook should always say: |
660 | |
661 | sub STORABLE_freeze { |
662 | my ($self, $cloning) = @_; |
663 | return if $cloning; # Regular default serialization |
664 | .... |
665 | } |
666 | |
667 | in order to keep reasonable dclone() semantics. |
668 | |
669 | =item C<STORABLE_thaw> I<obj>, I<cloning>, I<serialized>, ... |
670 | |
671 | The deserializing hook called on the object during deserialization. |
f062ea6c |
672 | But wait: if we're deserializing, there's no object yet... right? |
7a6a85bf |
673 | |
674 | Wrong: the Storable engine creates an empty one for you. If you know Eiffel, |
675 | you can view C<STORABLE_thaw> as an alternate creation routine. |
676 | |
677 | This means the hook can be inherited like any other method, and that |
678 | I<obj> is your blessed reference for this particular instance. |
679 | |
680 | The other arguments should look familiar if you know C<STORABLE_freeze>: |
681 | I<cloning> is true when we're part of a deep clone operation, I<serialized> |
682 | is the serialized string you returned to the engine in C<STORABLE_freeze>, |
683 | and there may be an optional list of references, in the same order you gave |
684 | them at serialization time, pointing to the deserialized objects (which |
685 | have been processed courtesy of the Storable engine). |
686 | |
212e9bde |
687 | When the Storable engine does not find any C<STORABLE_thaw> hook routine, |
688 | it tries to load the class by requiring the package dynamically (using |
689 | the blessed package name), and then re-attempts the lookup. If at that |
690 | time the hook cannot be located, the engine croaks. Note that this mechanism |
c261f00e |
691 | will fail if you define several classes in the same file, but L<perlmod> |
212e9bde |
692 | warned you. |
693 | |
f062ea6c |
694 | It is up to you to use this information to populate I<obj> the way you want. |
7a6a85bf |
695 | |
696 | Returned value: none. |
697 | |
698 | =back |
699 | |
700 | =head2 Predicates |
701 | |
c261f00e |
702 | Predicates are not exportable. They must be called by explicitly prefixing |
7a6a85bf |
703 | them with the Storable package name. |
704 | |
bbc7dcd2 |
705 | =over 4 |
7a6a85bf |
706 | |
707 | =item C<Storable::last_op_in_netorder> |
708 | |
709 | The C<Storable::last_op_in_netorder()> predicate will tell you whether |
710 | network order was used in the last store or retrieve operation. If you |
711 | don't know how to use this, just forget about it. |
712 | |
713 | =item C<Storable::is_storing> |
714 | |
715 | Returns true if within a store operation (via STORABLE_freeze hook). |
716 | |
717 | =item C<Storable::is_retrieving> |
718 | |
f062ea6c |
719 | Returns true if within a retrieve operation (via STORABLE_thaw hook). |
7a6a85bf |
720 | |
721 | =back |
722 | |
723 | =head2 Recursion |
724 | |
f062ea6c |
725 | With hooks comes the ability to recurse back to the Storable engine. |
726 | Indeed, hooks are regular Perl code, and Storable is convenient when |
727 | it comes to serializing and deserializing things, so why not use it |
728 | to handle the serialization string? |
7a6a85bf |
729 | |
f062ea6c |
730 | There are a few things you need to know, however: |
7a6a85bf |
731 | |
bbc7dcd2 |
732 | =over 4 |
7a6a85bf |
733 | |
734 | =item * |
735 | |
736 | You can create endless loops if the things you serialize via freeze() |
f062ea6c |
737 | (for instance) point back to the object we're trying to serialize in |
738 | the hook. |
7a6a85bf |
739 | |
740 | =item * |
741 | |
742 | Shared references among objects will not stay shared: if we're serializing |
743 | the list of object [A, C] where both object A and C refer to the SAME object |
744 | B, and if there is a serializing hook in A that says freeze(B), then when |
745 | deserializing, we'll get [A', C'] where A' refers to B', but C' refers to D, |
746 | a deep clone of B'. The topology was not preserved. |
747 | |
748 | =back |
749 | |
750 | That's why C<STORABLE_freeze> lets you provide a list of references |
751 | to serialize. The engine guarantees that those will be serialized in the |
752 | same context as the other objects, and therefore that shared objects will |
753 | stay shared. |
754 | |
755 | In the above [A, C] example, the C<STORABLE_freeze> hook could return: |
756 | |
757 | ("something", $self->{B}) |
758 | |
759 | and the B part would be serialized by the engine. In C<STORABLE_thaw>, you |
760 | would get back the reference to the B' object, deserialized for you. |
761 | |
762 | Therefore, recursion should normally be avoided, but is nonetheless supported. |
763 | |
764 | =head2 Deep Cloning |
765 | |
f062ea6c |
766 | There is a Clone module available on CPAN which implements deep cloning |
7a6a85bf |
767 | natively, i.e. without freezing to memory and thawing the result. It is |
768 | aimed to replace Storable's dclone() some day. However, it does not currently |
769 | support Storable hooks to redefine the way deep cloning is performed. |
770 | |
0a0da639 |
771 | =head1 Storable magic |
772 | |
773 | Yes, there's a lot of that :-) But more precisely, in UNIX systems |
774 | there's a utility called C<file>, which recognizes data files based on |
775 | their contents (usually their first few bytes). For this to work, |
8b793558 |
776 | a certain file called F<magic> needs to taught about the I<signature> |
0a0da639 |
777 | of the data. Where that configuration file lives depends on the UNIX |
f062ea6c |
778 | flavour; often it's something like F</usr/share/misc/magic> or |
8b793558 |
779 | F</etc/magic>. Your system administrator needs to do the updating of |
780 | the F<magic> file. The necessary signature information is output to |
f062ea6c |
781 | STDOUT by invoking Storable::show_file_magic(). Note that the GNU |
782 | implementation of the C<file> utility, version 3.38 or later, |
783 | is expected to contain support for recognising Storable files |
784 | out-of-the-box, in addition to other kinds of Perl files. |
0a0da639 |
785 | |
7a6a85bf |
786 | =head1 EXAMPLES |
787 | |
788 | Here are some code samples showing a possible usage of Storable: |
789 | |
790 | use Storable qw(store retrieve freeze thaw dclone); |
791 | |
792 | %color = ('Blue' => 0.1, 'Red' => 0.8, 'Black' => 0, 'White' => 1); |
793 | |
2359510d |
794 | store(\%color, 'mycolors') or die "Can't store %a in mycolors!\n"; |
7a6a85bf |
795 | |
2359510d |
796 | $colref = retrieve('mycolors'); |
797 | die "Unable to retrieve from mycolors!\n" unless defined $colref; |
7a6a85bf |
798 | printf "Blue is still %lf\n", $colref->{'Blue'}; |
799 | |
800 | $colref2 = dclone(\%color); |
801 | |
802 | $str = freeze(\%color); |
803 | printf "Serialization of %%color is %d bytes long.\n", length($str); |
804 | $colref3 = thaw($str); |
805 | |
806 | which prints (on my machine): |
807 | |
808 | Blue is still 0.100000 |
809 | Serialization of %color is 102 bytes long. |
810 | |
d2b96869 |
811 | Serialization of CODE references and deserialization in a safe |
812 | compartment: |
813 | |
197b90bc |
814 | =for example begin |
815 | |
d2b96869 |
816 | use Storable qw(freeze thaw); |
817 | use Safe; |
818 | use strict; |
819 | my $safe = new Safe; |
197b90bc |
820 | # because of opcodes used in "use strict": |
d1e2299c |
821 | $safe->permit(qw(:default require)); |
d2b96869 |
822 | local $Storable::Deparse = 1; |
823 | local $Storable::Eval = sub { $safe->reval($_[0]) }; |
197b90bc |
824 | my $serialized = freeze(sub { 42 }); |
d2b96869 |
825 | my $code = thaw($serialized); |
197b90bc |
826 | $code->() == 42; |
827 | |
828 | =for example end |
829 | |
830 | =for example_testing |
831 | is( $code->(), 42 ); |
d2b96869 |
832 | |
7a6a85bf |
833 | =head1 WARNING |
834 | |
835 | If you're using references as keys within your hash tables, you're bound |
f062ea6c |
836 | to be disappointed when retrieving your data. Indeed, Perl stringifies |
7a6a85bf |
837 | references used as hash table keys. If you later wish to access the |
838 | items via another reference stringification (i.e. using the same |
839 | reference that was used for the key originally to record the value into |
840 | the hash table), it will work because both references stringify to the |
841 | same string. |
842 | |
6fe6778b |
843 | It won't work across a sequence of C<store> and C<retrieve> operations, |
844 | however, because the addresses in the retrieved objects, which are |
845 | part of the stringified references, will probably differ from the |
846 | original addresses. The topology of your structure is preserved, |
847 | but not hidden semantics like those. |
7a6a85bf |
848 | |
849 | On platforms where it matters, be sure to call C<binmode()> on the |
850 | descriptors that you pass to Storable functions. |
851 | |
852 | Storing data canonically that contains large hashes can be |
853 | significantly slower than storing the same data normally, as |
c261f00e |
854 | temporary arrays to hold the keys for each hash have to be allocated, |
7a6a85bf |
855 | populated, sorted and freed. Some tests have shown a halving of the |
856 | speed of storing -- the exact penalty will depend on the complexity of |
857 | your data. There is no slowdown on retrieval. |
858 | |
859 | =head1 BUGS |
860 | |
197b90bc |
861 | You can't store GLOB, FORMLINE, etc.... If you can define semantics |
862 | for those operations, feel free to enhance Storable so that it can |
863 | deal with them. |
7a6a85bf |
864 | |
865 | The store functions will C<croak> if they run into such references |
866 | unless you set C<$Storable::forgive_me> to some C<TRUE> value. In that |
867 | case, the fatal message is turned in a warning and some |
868 | meaningless string is stored instead. |
869 | |
870 | Setting C<$Storable::canonical> may not yield frozen strings that |
871 | compare equal due to possible stringification of numbers. When the |
f062ea6c |
872 | string version of a scalar exists, it is the form stored; therefore, |
7a6a85bf |
873 | if you happen to use your numbers as strings between two freezing |
874 | operations on the same data structures, you will get different |
875 | results. |
876 | |
dd19458b |
877 | When storing doubles in network order, their value is stored as text. |
878 | However, you should also not expect non-numeric floating-point values |
879 | such as infinity and "not a number" to pass successfully through a |
880 | nstore()/retrieve() pair. |
881 | |
882 | As Storable neither knows nor cares about character sets (although it |
883 | does know that characters may be more than eight bits wide), any difference |
884 | in the interpretation of character codes between a host and a target |
885 | system is your problem. In particular, if host and target use different |
886 | code points to represent the characters used in the text representation |
887 | of floating-point numbers, you will not be able be able to exchange |
888 | floating-point data, even with nstore(). |
889 | |
c261f00e |
890 | C<Storable::drop_utf8> is a blunt tool. There is no facility either to |
891 | return B<all> strings as utf8 sequences, or to attempt to convert utf8 |
892 | data back to 8 bit and C<croak()> if the conversion fails. |
893 | |
ee0f7aac |
894 | Prior to Storable 2.01, no distinction was made between signed and |
895 | unsigned integers on storing. By default Storable prefers to store a |
896 | scalars string representation (if it has one) so this would only cause |
897 | problems when storing large unsigned integers that had never been coverted |
898 | to string or floating point. In other words values that had been generated |
899 | by integer operations such as logic ops and then not used in any string or |
900 | arithmetic context before storing. |
901 | |
902 | =head2 64 bit data in perl 5.6.0 and 5.6.1 |
903 | |
904 | This section only applies to you if you have existing data written out |
905 | by Storable 2.02 or earlier on perl 5.6.0 or 5.6.1 on Unix or Linux which |
906 | has been configured with 64 bit integer support (not the default) |
907 | If you got a precompiled perl, rather than running Configure to build |
908 | your own perl from source, then it almost certainly does not affect you, |
909 | and you can stop reading now (unless you're curious). If you're using perl |
910 | on Windows it does not affect you. |
911 | |
912 | Storable writes a file header which contains the sizes of various C |
913 | language types for the C compiler that built Storable (when not writing in |
914 | network order), and will refuse to load files written by a Storable not |
915 | on the same (or compatible) architecture. This check and a check on |
916 | machine byteorder is needed because the size of various fields in the file |
917 | are given by the sizes of the C language types, and so files written on |
918 | different architectures are incompatible. This is done for increased speed. |
919 | (When writing in network order, all fields are written out as standard |
920 | lengths, which allows full interworking, but takes longer to read and write) |
921 | |
922 | Perl 5.6.x introduced the ability to optional configure the perl interpreter |
923 | to use C's C<long long> type to allow scalars to store 64 bit integers on 32 |
924 | bit systems. However, due to the way the Perl configuration system |
925 | generated the C configuration files on non-Windows platforms, and the way |
926 | Storable generates its header, nothing in the Storable file header reflected |
927 | whether the perl writing was using 32 or 64 bit integers, despite the fact |
928 | that Storable was storing some data differently in the file. Hence Storable |
929 | running on perl with 64 bit integers will read the header from a file |
930 | written by a 32 bit perl, not realise that the data is actually in a subtly |
931 | incompatible format, and then go horribly wrong (possibly crashing) if it |
932 | encountered a stored integer. This is a design failure. |
933 | |
934 | Storable has now been changed to write out and read in a file header with |
935 | information about the size of integers. It's impossible to detect whether |
936 | an old file being read in was written with 32 or 64 bit integers (they have |
937 | the same header) so it's impossible to automatically switch to a correct |
938 | backwards compatibility mode. Hence this Storable defaults to the new, |
939 | correct behaviour. |
940 | |
941 | What this means is that if you have data written by Storable 1.x running |
942 | on perl 5.6.0 or 5.6.1 configured with 64 bit integers on Unix or Linux |
943 | then by default this Storable will refuse to read it, giving the error |
944 | I<Byte order is not compatible>. If you have such data then you you |
945 | should set C<$Storable::interwork_56_64bit> to a true value to make this |
946 | Storable read and write files with the old header. You should also |
947 | migrate your data, or any older perl you are communicating with, to this |
948 | current version of Storable. |
949 | |
950 | If you don't have data written with specific configuration of perl described |
951 | above, then you do not and should not do anything. Don't set the flag - |
952 | not only will Storable on an identically configured perl refuse to load them, |
953 | but Storable a differently configured perl will load them believing them |
954 | to be correct for it, and then may well fail or crash part way through |
955 | reading them. |
956 | |
7a6a85bf |
957 | =head1 CREDITS |
958 | |
959 | Thank you to (in chronological order): |
960 | |
961 | Jarkko Hietaniemi <jhi@iki.fi> |
962 | Ulrich Pfeifer <pfeifer@charly.informatik.uni-dortmund.de> |
963 | Benjamin A. Holzman <bah@ecnvantage.com> |
964 | Andrew Ford <A.Ford@ford-mason.co.uk> |
965 | Gisle Aas <gisle@aas.no> |
966 | Jeff Gresham <gresham_jeffrey@jpmorgan.com> |
967 | Murray Nesbitt <murray@activestate.com> |
968 | Marc Lehmann <pcg@opengroup.org> |
9e21b3d0 |
969 | Justin Banks <justinb@wamnet.com> |
970 | Jarkko Hietaniemi <jhi@iki.fi> (AGAIN, as perl 5.7.0 Pumpkin!) |
dd19458b |
971 | Salvador Ortiz Garcia <sog@msg.com.mx> |
972 | Dominic Dunlop <domo@computer.org> |
973 | Erik Haugan <erik@solbors.no> |
7a6a85bf |
974 | |
975 | for their bug reports, suggestions and contributions. |
976 | |
977 | Benjamin Holzman contributed the tied variable support, Andrew Ford |
978 | contributed the canonical order for hashes, and Gisle Aas fixed |
f062ea6c |
979 | a few misunderstandings of mine regarding the perl internals, |
7a6a85bf |
980 | and optimized the emission of "tags" in the output streams by |
981 | simply counting the objects instead of tagging them (leading to |
982 | a binary incompatibility for the Storable image starting at version |
f062ea6c |
983 | 0.6--older images are, of course, still properly understood). |
7a6a85bf |
984 | Murray Nesbitt made Storable thread-safe. Marc Lehmann added overloading |
f062ea6c |
985 | and references to tied items support. |
7a6a85bf |
986 | |
7a6a85bf |
987 | =head1 AUTHOR |
988 | |
0ba8809e |
989 | Storable was written by Raphael Manfredi F<E<lt>Raphael_Manfredi@pobox.comE<gt>> |
775ecd75 |
990 | Maintenance is now done by the perl5-porters F<E<lt>perl5-porters@perl.orgE<gt>> |
0ba8809e |
991 | |
992 | Please e-mail us with problems, bug fixes, comments and complaints, |
993 | although if you have complements you should send them to Raphael. |
994 | Please don't e-mail Raphael with problems, as he no longer works on |
995 | Storable, and your message will be delayed while he forwards it to us. |
7a6a85bf |
996 | |
997 | =head1 SEE ALSO |
998 | |
c261f00e |
999 | L<Clone>. |
7a6a85bf |
1000 | |
1001 | =cut |