ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/CBOR-XS/XS.pm

Comparing CBOR-XS/XS.pm (file contents):
Revision 1.1 by root, Fri Oct 25 23:09:45 2013 UTC vs.
Revision 1.78 by root, Tue Dec 8 08:29:44 2020 UTC

…		…
12	$perl_value = decode_cbor $binary_cbor_data;	12	$perl_value = decode_cbor $binary_cbor_data;
13		13
14	# OO-interface	14	# OO-interface
15		15
16	$coder = CBOR::XS->new;	16	$coder = CBOR::XS->new;
17	#TODO	17	$binary_cbor_data = $coder->encode ($perl_value);
		18	$perl_value = $coder->decode ($binary_cbor_data);
		19
		20	# prefix decoding
		21
		22	my $many_cbor_strings = ...;
		23	while (length $many_cbor_strings) {
		24	my ($data, $length) = $cbor->decode_prefix ($many_cbor_strings);
		25	# data was decoded
		26	substr $many_cbor_strings, 0, $length, ""; # remove decoded cbor string
		27	}
18		28
19	=head1 DESCRIPTION	29	=head1 DESCRIPTION
20		30
21	WARNING! THIS IS A PRE-ALPHA RELEASE! IT WILL CRASH, CORRUPT YOUR DATA AND	31	This module converts Perl data structures to the Concise Binary Object
22	EAT YOUR CHILDREN!	32	Representation (CBOR) and vice versa. CBOR is a fast binary serialisation
		33	format that aims to use an (almost) superset of the JSON data model, i.e.
		34	when you can represent something useful in JSON, you should be able to
		35	represent it in CBOR.
23		36
24	This module converts Perl data structures to CBOR and vice versa. Its	37	In short, CBOR is a faster and quite compact binary alternative to JSON,
		38	with the added ability of supporting serialisation of Perl objects. (JSON
		39	often compresses better than CBOR though, so if you plan to compress the
		40	data later and speed is less important you might want to compare both
		41	formats first).
		42
25	primary goal is to be I<correct> and its secondary goal is to be	43	The primary goal of this module is to be I<correct> and the secondary goal
26	I<fast>. To reach the latter goal it was written in C.	44	is to be I<fast>. To reach the latter goal it was written in C.
		45
		46	To give you a general idea about speed, with texts in the megabyte range,
		47	C<CBOR::XS> usually encodes roughly twice as fast as L<Storable> or
		48	L<JSON::XS> and decodes about 15%-30% faster than those. The shorter the
		49	data, the worse L<Storable> performs in comparison.
		50
		51	Regarding compactness, C<CBOR::XS>-encoded data structures are usually
		52	about 20% smaller than the same data encoded as (compact) JSON or
		53	L<Storable>.
		54
		55	In addition to the core CBOR data format, this module implements a
		56	number of extensions, to support cyclic and shared data structures
		57	(see C<allow_sharing> and C<allow_cycles>), string deduplication (see
		58	C<pack_strings>) and scalar references (always enabled).
27		59
28	See MAPPING, below, on how CBOR::XS maps perl values to CBOR values and	60	See MAPPING, below, on how CBOR::XS maps perl values to CBOR values and
29	vice versa.	61	vice versa.
30		62
31	=cut	63	=cut
32		64
33	package CBOR::XS;	65	package CBOR::XS;
34		66
35	use common::sense;	67	use common::sense;
36		68
37	our $VERSION = 0.01;	69	our $VERSION = 1.83;
38	our @ISA = qw(Exporter);	70	our @ISA = qw(Exporter);
39		71
40	our @EXPORT = qw(encode_cbor decode_cbor);	72	our @EXPORT = qw(encode_cbor decode_cbor);
41		73
42	use Exporter;	74	use Exporter;
43	use XSLoader;	75	use XSLoader;
44		76
		77	use Types::Serialiser;
		78
		79	our $MAGIC = "\xd9\xd9\xf7";
		80
45	=head1 FUNCTIONAL INTERFACE	81	=head1 FUNCTIONAL INTERFACE
46		82
47	The following convenience methods are provided by this module. They are	83	The following convenience methods are provided by this module. They are
48	exported by default:	84	exported by default:
49		85
…		…
75	strings. All boolean flags described below are by default I<disabled>.	111	strings. All boolean flags described below are by default I<disabled>.
76		112
77	The mutators for flags all return the CBOR object again and thus calls can	113	The mutators for flags all return the CBOR object again and thus calls can
78	be chained:	114	be chained:
79		115
80	#TODO
81	my $cbor = CBOR::XS->new->encode ({a => [1,2]});	116	my $cbor = CBOR::XS->new->encode ({a => [1,2]});
		117
		118	=item $cbor = new_safe CBOR::XS
		119
		120	Create a new, safe/secure CBOR::XS object. This is similar to C<new>,
		121	but configures the coder object to be safe to use with untrusted
		122	data. Currently, this is equivalent to:
		123
		124	my $cbor = CBOR::XS
		125	->new
		126	->forbid_objects
		127	->filter (\&CBOR::XS::safe_filter)
		128	->max_size (1e8);
		129
		130	But is more future proof (it is better to crash because of a change than
		131	to be exploited in other ways).
		132
		133	=cut
		134
		135	sub new_safe {
		136	CBOR::XS
		137	->new
		138	->forbid_objects
		139	->filter (\&CBOR::XS::safe_filter)
		140	->max_size (1e8)
		141	}
82		142
83	=item $cbor = $cbor->max_depth ([$maximum_nesting_depth])	143	=item $cbor = $cbor->max_depth ([$maximum_nesting_depth])
84		144
85	=item $max_depth = $cbor->get_max_depth	145	=item $max_depth = $cbor->get_max_depth
86		146
…		…
102		162
103	Note that nesting is implemented by recursion in C. The default value has	163	Note that nesting is implemented by recursion in C. The default value has
104	been chosen to be as large as typical operating systems allow without	164	been chosen to be as large as typical operating systems allow without
105	crashing.	165	crashing.
106		166
107	See SECURITY CONSIDERATIONS, below, for more info on why this is useful.	167	See L<SECURITY CONSIDERATIONS>, below, for more info on why this is useful.
108		168
109	=item $cbor = $cbor->max_size ([$maximum_string_size])	169	=item $cbor = $cbor->max_size ([$maximum_string_size])
110		170
111	=item $max_size = $cbor->get_max_size	171	=item $max_size = $cbor->get_max_size
112		172
…		…
117	effect on C<encode> (yet).	177	effect on C<encode> (yet).
118		178
119	If no argument is given, the limit check will be deactivated (same as when	179	If no argument is given, the limit check will be deactivated (same as when
120	C<0> is specified).	180	C<0> is specified).
121		181
122	See SECURITY CONSIDERATIONS, below, for more info on why this is useful.	182	See L<SECURITY CONSIDERATIONS>, below, for more info on why this is useful.
		183
		184	=item $cbor = $cbor->allow_unknown ([$enable])
		185
		186	=item $enabled = $cbor->get_allow_unknown
		187
		188	If C<$enable> is true (or missing), then C<encode> will I<not> throw an
		189	exception when it encounters values it cannot represent in CBOR (for
		190	example, filehandles) but instead will encode a CBOR C<error> value.
		191
		192	If C<$enable> is false (the default), then C<encode> will throw an
		193	exception when it encounters anything it cannot encode as CBOR.
		194
		195	This option does not affect C<decode> in any way, and it is recommended to
		196	leave it off unless you know your communications partner.
		197
		198	=item $cbor = $cbor->allow_sharing ([$enable])
		199
		200	=item $enabled = $cbor->get_allow_sharing
		201
		202	If C<$enable> is true (or missing), then C<encode> will not double-encode
		203	values that have been referenced before (e.g. when the same object, such
		204	as an array, is referenced multiple times), but instead will emit a
		205	reference to the earlier value.
		206
		207	This means that such values will only be encoded once, and will not result
		208	in a deep cloning of the value on decode, in decoders supporting the value
		209	sharing extension. This also makes it possible to encode cyclic data
		210	structures (which need C<allow_cycles> to be enabled to be decoded by this
		211	module).
		212
		213	It is recommended to leave it off unless you know your
		214	communication partner supports the value sharing extensions to CBOR
		215	(L<http://cbor.schmorp.de/value-sharing>), as without decoder support, the
		216	resulting data structure might be unusable.
		217
		218	Detecting shared values incurs a runtime overhead when values are encoded
		219	that have a reference counter large than one, and might unnecessarily
		220	increase the encoded size, as potentially shared values are encoded as
		221	shareable whether or not they are actually shared.
		222
		223	At the moment, only targets of references can be shared (e.g. scalars,
		224	arrays or hashes pointed to by a reference). Weirder constructs, such as
		225	an array with multiple "copies" of the I<same> string, which are hard but
		226	not impossible to create in Perl, are not supported (this is the same as
		227	with L<Storable>).
		228
		229	If C<$enable> is false (the default), then C<encode> will encode shared
		230	data structures repeatedly, unsharing them in the process. Cyclic data
		231	structures cannot be encoded in this mode.
		232
		233	This option does not affect C<decode> in any way - shared values and
		234	references will always be decoded properly if present.
		235
		236	=item $cbor = $cbor->allow_cycles ([$enable])
		237
		238	=item $enabled = $cbor->get_allow_cycles
		239
		240	If C<$enable> is true (or missing), then C<decode> will happily decode
		241	self-referential (cyclic) data structures. By default these will not be
		242	decoded, as they need manual cleanup to avoid memory leaks, so code that
		243	isn't prepared for this will not leak memory.
		244
		245	If C<$enable> is false (the default), then C<decode> will throw an error
		246	when it encounters a self-referential/cyclic data structure.
		247
		248	FUTURE DIRECTION: the motivation behind this option is to avoid I<real>
		249	cycles - future versions of this module might chose to decode cyclic data
		250	structures using weak references when this option is off, instead of
		251	throwing an error.
		252
		253	This option does not affect C<encode> in any way - shared values and
		254	references will always be encoded properly if present.
		255
		256	=item $cbor = $cbor->forbid_objects ([$enable])
		257
		258	=item $enabled = $cbor->get_forbid_objects
		259
		260	Disables the use of the object serialiser protocol.
		261
		262	If C<$enable> is true (or missing), then C<encode> will will throw an
		263	exception when it encounters perl objects that would be encoded using the
		264	perl-object tag (26). When C<decode> encounters such tags, it will fall
		265	back to the general filter/tagged logic as if this were an unknown tag (by
		266	default resulting in a C<CBOR::XC::Tagged> object).
		267
		268	If C<$enable> is false (the default), then C<encode> will use the
		269	L<Types::Serialiser> object serialisation protocol to serialise objects
		270	into perl-object tags, and C<decode> will do the same to decode such tags.
		271
		272	See L<SECURITY CONSIDERATIONS>, below, for more info on why forbidding this
		273	protocol can be useful.
		274
		275	=item $cbor = $cbor->pack_strings ([$enable])
		276
		277	=item $enabled = $cbor->get_pack_strings
		278
		279	If C<$enable> is true (or missing), then C<encode> will try not to encode
		280	the same string twice, but will instead encode a reference to the string
		281	instead. Depending on your data format, this can save a lot of space, but
		282	also results in a very large runtime overhead (expect encoding times to be
		283	2-4 times as high as without).
		284
		285	It is recommended to leave it off unless you know your
		286	communications partner supports the stringref extension to CBOR
		287	(L<http://cbor.schmorp.de/stringref>), as without decoder support, the
		288	resulting data structure might not be usable.
		289
		290	If C<$enable> is false (the default), then C<encode> will encode strings
		291	the standard CBOR way.
		292
		293	This option does not affect C<decode> in any way - string references will
		294	always be decoded properly if present.
		295
		296	=item $cbor = $cbor->text_keys ([$enable])
		297
		298	=item $enabled = $cbor->get_text_keys
		299
		300	If C<$enabled> is true (or missing), then C<encode> will encode all
		301	perl hash keys as CBOR text strings/UTF-8 string, upgrading them as needed.
		302
		303	If C<$enable> is false (the default), then C<encode> will encode hash keys
		304	normally - upgraded perl strings (strings internally encoded as UTF-8) as
		305	CBOR text strings, and downgraded perl strings as CBOR byte strings.
		306
		307	This option does not affect C<decode> in any way.
		308
		309	This option is useful for interoperability with CBOR decoders that don't
		310	treat byte strings as a form of text. It is especially useful as Perl
		311	gives very little control over hash keys.
		312
		313	Enabling this option can be slow, as all downgraded hash keys that are
		314	encoded need to be scanned and converted to UTF-8.
		315
		316	=item $cbor = $cbor->text_strings ([$enable])
		317
		318	=item $enabled = $cbor->get_text_strings
		319
		320	This option works similar to C<text_keys>, above, but works on all strings
		321	(including hash keys), so C<text_keys> has no further effect after
		322	enabling C<text_strings>.
		323
		324	If C<$enabled> is true (or missing), then C<encode> will encode all perl
		325	strings as CBOR text strings/UTF-8 strings, upgrading them as needed.
		326
		327	If C<$enable> is false (the default), then C<encode> will encode strings
		328	normally (but see C<text_keys>) - upgraded perl strings (strings
		329	internally encoded as UTF-8) as CBOR text strings, and downgraded perl
		330	strings as CBOR byte strings.
		331
		332	This option does not affect C<decode> in any way.
		333
		334	This option has similar advantages and disadvantages as C<text_keys>. In
		335	addition, this option effectively removes the ability to automatically
		336	encode byte strings, which might break some C<FREEZE> and C<TO_CBOR>
		337	methods that rely on this.
		338
		339	A workaround is to use explicit type casts, which are unaffected by this option.
		340
		341	=item $cbor = $cbor->validate_utf8 ([$enable])
		342
		343	=item $enabled = $cbor->get_validate_utf8
		344
		345	If C<$enable> is true (or missing), then C<decode> will validate that
		346	elements (text strings) containing UTF-8 data in fact contain valid UTF-8
		347	data (instead of blindly accepting it). This validation obviously takes
		348	extra time during decoding.
		349
		350	The concept of "valid UTF-8" used is perl's concept, which is a superset
		351	of the official UTF-8.
		352
		353	If C<$enable> is false (the default), then C<decode> will blindly accept
		354	UTF-8 data, marking them as valid UTF-8 in the resulting data structure
		355	regardless of whether that's true or not.
		356
		357	Perl isn't too happy about corrupted UTF-8 in strings, but should
		358	generally not crash or do similarly evil things. Extensions might be not
		359	so forgiving, so it's recommended to turn on this setting if you receive
		360	untrusted CBOR.
		361
		362	This option does not affect C<encode> in any way - strings that are
		363	supposedly valid UTF-8 will simply be dumped into the resulting CBOR
		364	string without checking whether that is, in fact, true or not.
		365
		366	=item $cbor = $cbor->filter ([$cb->($tag, $value)])
		367
		368	=item $cb_or_undef = $cbor->get_filter
		369
		370	Sets or replaces the tagged value decoding filter (when C<$cb> is
		371	specified) or clears the filter (if no argument or C<undef> is provided).
		372
		373	The filter callback is called only during decoding, when a non-enforced
		374	tagged value has been decoded (see L<TAG HANDLING AND EXTENSIONS> for a
		375	list of enforced tags). For specific tags, it's often better to provide a
		376	default converter using the C<%CBOR::XS::FILTER> hash (see below).
		377
		378	The first argument is the numerical tag, the second is the (decoded) value
		379	that has been tagged.
		380
		381	The filter function should return either exactly one value, which will
		382	replace the tagged value in the decoded data structure, or no values,
		383	which will result in default handling, which currently means the decoder
		384	creates a C<CBOR::XS::Tagged> object to hold the tag and the value.
		385
		386	When the filter is cleared (the default state), the default filter
		387	function, C<CBOR::XS::default_filter>, is used. This function simply
		388	looks up the tag in the C<%CBOR::XS::FILTER> hash. If an entry exists
		389	it must be a code reference that is called with tag and value, and is
		390	responsible for decoding the value. If no entry exists, it returns no
		391	values. C<CBOR::XS> provides a number of default filter functions already,
		392	the the C<%CBOR::XS::FILTER> hash can be freely extended with more.
		393
		394	C<CBOR::XS> additionally provides an alternative filter function that is
		395	supposed to be safe to use with untrusted data (which the default filter
		396	might not), called C<CBOR::XS::safe_filter>, which works the same as
		397	the C<default_filter> but uses the C<%CBOR::XS::SAFE_FILTER> variable
		398	instead. It is prepopulated with the tag decoding functions that are
		399	deemed safe (basically the same as C<%CBOR::XS::FILTER> without all
		400	the bignum tags), and can be extended by user code as wlel, although,
		401	obviously, one should be very careful about adding decoding functions
		402	here, since the expectation is that they are safe to use on untrusted
		403	data, after all.
		404
		405	Example: decode all tags not handled internally into C<CBOR::XS::Tagged>
		406	objects, with no other special handling (useful when working with
		407	potentially "unsafe" CBOR data).
		408
		409	CBOR::XS->new->filter (sub { })->decode ($cbor_data);
		410
		411	Example: provide a global filter for tag 1347375694, converting the value
		412	into some string form.
		413
		414	$CBOR::XS::FILTER{1347375694} = sub {
		415	my ($tag, $value);
		416
		417	"tag 1347375694 value $value"
		418	};
		419
		420	Example: provide your own filter function that looks up tags in your own
		421	hash:
		422
		423	my %my_filter = (
		424	998347484 => sub {
		425	my ($tag, $value);
		426
		427	"tag 998347484 value $value"
		428	};
		429	);
		430
		431	my $coder = CBOR::XS->new->filter (sub {
		432	&{ $my_filter{$_[0]} or return }
		433	});
		434
		435
		436	Example: use the safe filter function (see L<SECURITY CONSIDERATIONS> for
		437	more considerations on security).
		438
		439	CBOR::XS->new->filter (\&CBOR::XS::safe_filter)->decode ($cbor_data);
123		440
124	=item $cbor_data = $cbor->encode ($perl_scalar)	441	=item $cbor_data = $cbor->encode ($perl_scalar)
125		442
126	Converts the given Perl data structure (a scalar value) to its CBOR	443	Converts the given Perl data structure (a scalar value) to its CBOR
127	representation.	444	representation.
…		…
137	when there is trailing garbage after the CBOR string, it will silently	454	when there is trailing garbage after the CBOR string, it will silently
138	stop parsing there and return the number of characters consumed so far.	455	stop parsing there and return the number of characters consumed so far.
139		456
140	This is useful if your CBOR texts are not delimited by an outer protocol	457	This is useful if your CBOR texts are not delimited by an outer protocol
141	and you need to know where the first CBOR string ends amd the next one	458	and you need to know where the first CBOR string ends amd the next one
142	starts.	459	starts - CBOR strings are self-delimited, so it is possible to concatenate
		460	CBOR strings without any delimiters or size fields and recover their data.
143		461
144	CBOR::XS->new->decode_prefix ("......")	462	CBOR::XS->new->decode_prefix ("......")
145	=> ("...", 3)	463	=> ("...", 3)
		464
		465	=back
		466
		467	=head2 INCREMENTAL PARSING
		468
		469	In some cases, there is the need for incremental parsing of JSON
		470	texts. While this module always has to keep both CBOR text and resulting
		471	Perl data structure in memory at one time, it does allow you to parse a
		472	CBOR stream incrementally, using a similar to using "decode_prefix" to see
		473	if a full CBOR object is available, but is much more efficient.
		474
		475	It basically works by parsing as much of a CBOR string as possible - if
		476	the CBOR data is not complete yet, the pasrer will remember where it was,
		477	to be able to restart when more data has been accumulated. Once enough
		478	data is available to either decode a complete CBOR value or raise an
		479	error, a real decode will be attempted.
		480
		481	A typical use case would be a network protocol that consists of sending
		482	and receiving CBOR-encoded messages. The solution that works with CBOR and
		483	about anything else is by prepending a length to every CBOR value, so the
		484	receiver knows how many octets to read. More compact (and slightly slower)
		485	would be to just send CBOR values back-to-back, as C<CBOR::XS> knows where
		486	a CBOR value ends, and doesn't need an explicit length.
		487
		488	The following methods help with this:
		489
		490	=over 4
		491
		492	=item @decoded = $cbor->incr_parse ($buffer)
		493
		494	This method attempts to decode exactly one CBOR value from the beginning
		495	of the given C<$buffer>. The value is removed from the C<$buffer> on
		496	success. When C<$buffer> doesn't contain a complete value yet, it returns
		497	nothing. Finally, when the C<$buffer> doesn't start with something
		498	that could ever be a valid CBOR value, it raises an exception, just as
		499	C<decode> would. In the latter case the decoder state is undefined and
		500	must be reset before being able to parse further.
		501
		502	This method modifies the C<$buffer> in place. When no CBOR value can be
		503	decoded, the decoder stores the current string offset. On the next call,
		504	continues decoding at the place where it stopped before. For this to make
		505	sense, the C<$buffer> must begin with the same octets as on previous
		506	unsuccessful calls.
		507
		508	You can call this method in scalar context, in which case it either
		509	returns a decoded value or C<undef>. This makes it impossible to
		510	distinguish between CBOR null values (which decode to C<undef>) and an
		511	unsuccessful decode, which is often acceptable.
		512
		513	=item @decoded = $cbor->incr_parse_multiple ($buffer)
		514
		515	Same as C<incr_parse>, but attempts to decode as many CBOR values as
		516	possible in one go, instead of at most one. Calls to C<incr_parse> and
		517	C<incr_parse_multiple> can be interleaved.
		518
		519	=item $cbor->incr_reset
		520
		521	Resets the incremental decoder. This throws away any saved state, so that
		522	subsequent calls to C<incr_parse> or C<incr_parse_multiple> start to parse
		523	a new CBOR value from the beginning of the C<$buffer> again.
		524
		525	This method can be called at any time, but it I<must> be called if you want
		526	to change your C<$buffer> or there was a decoding error and you want to
		527	reuse the C<$cbor> object for future incremental parsings.
146		528
147	=back	529	=back
148		530
149		531
150	=head1 MAPPING	532	=head1 MAPPING
…		…
161		543
162	=head2 CBOR -> PERL	544	=head2 CBOR -> PERL
163		545
164	=over 4	546	=over 4
165		547
166	=item True, False	548	=item integers
167		549
168	These CBOR values become C<CBOR::XS::true> and C<CBOR::XS::false>,	550	CBOR integers become (numeric) perl scalars. On perls without 64 bit
		551	support, 64 bit integers will be truncated or otherwise corrupted.
		552
		553	=item byte strings
		554
		555	Byte strings will become octet strings in Perl (the Byte values 0..255
		556	will simply become characters of the same value in Perl).
		557
		558	=item UTF-8 strings
		559
		560	UTF-8 strings in CBOR will be decoded, i.e. the UTF-8 octets will be
		561	decoded into proper Unicode code points. At the moment, the validity of
		562	the UTF-8 octets will not be validated - corrupt input will result in
		563	corrupted Perl strings.
		564
		565	=item arrays, maps
		566
		567	CBOR arrays and CBOR maps will be converted into references to a Perl
		568	array or hash, respectively. The keys of the map will be stringified
		569	during this process.
		570
		571	=item null
		572
		573	CBOR null becomes C<undef> in Perl.
		574
		575	=item true, false, undefined
		576
		577	These CBOR values become C<Types:Serialiser::true>,
		578	C<Types:Serialiser::false> and C<Types::Serialiser::error>,
169	respectively. They are overloaded to act almost exactly like the numbers	579	respectively. They are overloaded to act almost exactly like the numbers
170	C<1> and C<0>. You can check whether a scalar is a CBOR boolean by using	580	C<1> and C<0> (for true and false) or to throw an exception on access (for
171	the C<CBOR::XS::is_bool> function.	581	error). See the L<Types::Serialiser> manpage for details.
172		582
173	=item null	583	=item tagged values
174		584
175	A CBOR Null value becomes C<undef> in Perl.	585	Tagged items consists of a numeric tag and another CBOR value.
		586
		587	See L<TAG HANDLING AND EXTENSIONS> and the description of C<< ->filter >>
		588	for details on which tags are handled how.
		589
		590	=item anything else
		591
		592	Anything else (e.g. unsupported simple values) will raise a decoding
		593	error.
176		594
177	=back	595	=back
178		596
179		597
180	=head2 PERL -> CBOR	598	=head2 PERL -> CBOR
181		599
182	The mapping from Perl to CBOR is slightly more difficult, as Perl is a	600	The mapping from Perl to CBOR is slightly more difficult, as Perl is a
183	truly typeless language, so we can only guess which CBOR type is meant by	601	typeless language. That means this module can only guess which CBOR type
184	a Perl value.	602	is meant by a perl value.
185		603
186	=over 4	604	=over 4
187		605
188	=item hash references	606	=item hash references
189		607
190	Perl hash references become CBOR maps. As there is no inherent ordering	608	Perl hash references become CBOR maps. As there is no inherent ordering in
191	in hash keys (or CBOR maps), they will usually be encoded in a	609	hash keys (or CBOR maps), they will usually be encoded in a pseudo-random
192	pseudo-random order.	610	order. This order can be different each time a hash is encoded.
		611
		612	Currently, tied hashes will use the indefinite-length format, while normal
		613	hashes will use the fixed-length format.
193		614
194	=item array references	615	=item array references
195		616
196	Perl array references become CBOR arrays.	617	Perl array references become fixed-length CBOR arrays.
197		618
198	=item other references	619	=item other references
199		620
200	Other unblessed references are generally not allowed and will cause an	621	Other unblessed references will be represented using
201	exception to be thrown, except for references to the integers C<0> and	622	the indirection tag extension (tag value C<22098>,
202	C<1>, which get turned into C<False> and C<True> in CBOR.	623	L<http://cbor.schmorp.de/indirection>). CBOR decoders are guaranteed
		624	to be able to decode these values somehow, by either "doing the right
		625	thing", decoding into a generic tagged object, simply ignoring the tag, or
		626	something else.
203		627
204	=item CBOR::XS::true, CBOR::XS::false	628	=item CBOR::XS::Tagged objects
205		629
		630	Objects of this type must be arrays consisting of a single C<[tag, value]>
		631	pair. The (numerical) tag will be encoded as a CBOR tag, the value will
		632	be encoded as appropriate for the value. You must use C<CBOR::XS::tag> to
		633	create such objects.
		634
		635	=item Types::Serialiser::true, Types::Serialiser::false, Types::Serialiser::error
		636
206	These special values become CBOR True and CBOR False values,	637	These special values become CBOR true, CBOR false and CBOR undefined
207	respectively. You can also use C<\1> and C<\0> directly if you want.	638	values, respectively. You can also use C<\1>, C<\0> and C<\undef> directly
		639	if you want.
208		640
209	=item blessed objects	641	=item other blessed objects
210		642
211	Blessed objects are not directly representable in CBOR. TODO	643	Other blessed objects are serialised via C<TO_CBOR> or C<FREEZE>. See
212	See the	644	L<TAG HANDLING AND EXTENSIONS> for specific classes handled by this
213	C<allow_blessed> and C<convert_blessed> methods on various options on	645	module, and L<OBJECT SERIALISATION> for generic object serialisation.
214	how to deal with this: basically, you can choose between throwing an
215	exception, encoding the reference as if it weren't blessed, or provide
216	your own serialiser method.
217		646
218	=item simple scalars	647	=item simple scalars
219		648
220	TODO
221	Simple Perl scalars (any scalar that is not a reference) are the most	649	Simple Perl scalars (any scalar that is not a reference) are the most
222	difficult objects to encode: CBOR::XS will encode undefined scalars as	650	difficult objects to encode: CBOR::XS will encode undefined scalars as
223	CBOR C<Null> values, scalars that have last been used in a string context	651	CBOR null values, scalars that have last been used in a string context
224	before encoding as CBOR strings, and anything else as number value:	652	before encoding as CBOR strings, and anything else as number value:
225		653
226	# dump as number	654	# dump as number
227	encode_cbor [2] # yields [2]	655	encode_cbor [2] # yields [2]
228	encode_cbor [-3.0e17] # yields [-3e+17]	656	encode_cbor [-3.0e17] # yields [-3e+17]
229	my $value = 5; encode_cbor [$value] # yields [5]	657	my $value = 5; encode_cbor [$value] # yields [5]
230		658
231	# used as string, so dump as string	659	# used as string, so dump as string (either byte or text)
232	print $value;	660	print $value;
233	encode_cbor [$value] # yields ["5"]	661	encode_cbor [$value] # yields ["5"]
234		662
235	# undef becomes null	663	# undef becomes null
236	encode_cbor [undef] # yields [null]	664	encode_cbor [undef] # yields [null]
…		…
240	my $x = 3.1; # some variable containing a number	668	my $x = 3.1; # some variable containing a number
241	"$x"; # stringified	669	"$x"; # stringified
242	$x .= ""; # another, more awkward way to stringify	670	$x .= ""; # another, more awkward way to stringify
243	print $x; # perl does it for you, too, quite often	671	print $x; # perl does it for you, too, quite often
244		672
		673	You can force whether a string is encoded as byte or text string by using
		674	C<utf8::upgrade> and C<utf8::downgrade> (if C<text_strings> is disabled).
		675
		676	utf8::upgrade $x; # encode $x as text string
		677	utf8::downgrade $x; # encode $x as byte string
		678
		679	More options are available, see L<TYPE CASTS>, below, and the C<text_keys>
		680	and C<text_strings> options.
		681
		682	Perl doesn't define what operations up- and downgrade strings, so if the
		683	difference between byte and text is important, you should up- or downgrade
		684	your string as late as possible before encoding. You can also force the
		685	use of CBOR text strings by using C<text_keys> or C<text_strings>.
		686
245	You can force the type to be a CBOR number by numifying it:	687	You can force the type to be a CBOR number by numifying it:
246		688
247	my $x = "3"; # some variable containing a string	689	my $x = "3"; # some variable containing a string
248	$x += 0; # numify it, ensuring it will be dumped as a number	690	$x += 0; # numify it, ensuring it will be dumped as a number
249	$x *= 1; # same thing, the choice is yours.	691	$x *= 1; # same thing, the choice is yours.
250		692
251	You can not currently force the type in other, less obscure, ways. Tell me	693	You can not currently force the type in other, less obscure, ways. Tell me
252	if you need this capability (but don't forget to explain why it's needed	694	if you need this capability (but don't forget to explain why it's needed
253	:).	695	:).
254		696
255	Note that numerical precision has the same meaning as under Perl (so	697	Perl values that seem to be integers generally use the shortest possible
256	binary to decimal conversion follows the same rules as in Perl, which	698	representation. Floating-point values will use either the IEEE single
257	can differ to other languages). Also, your perl interpreter might expose	699	format if possible without loss of precision, otherwise the IEEE double
258	extensions to the floating point numbers of your platform, such as	700	format will be used. Perls that use formats other than IEEE double to
259	infinities or NaN's - these cannot be represented in CBOR, and it is an	701	represent numerical values are supported, but might suffer loss of
260	error to pass those in.	702	precision.
261		703
262	=back	704	=back
263		705
		706	=head2 TYPE CASTS
264		707
		708	B<EXPERIMENTAL>: As an experimental extension, C<CBOR::XS> allows you to
		709	force specific cbor types to be used when encoding. That allows you to
		710	encode types not normally accessible (e.g. half floats) as well as force
		711	string types even when C<text_strings> is in effect.
		712
		713	Type forcing is done by calling a special "cast" function which keeps a
		714	copy of the value and returns a new value that can be handed over to any
		715	CBOR encoder function.
		716
		717	The following casts are currently available (all of which are unary operators):
		718
		719	=over
		720
		721	=item CBOR::XS::as_int $value
		722
		723	Forces the value to be encoded as some form of (basic, not bignum) integer
		724	type.
		725
		726	=item CBOR::XS::as_text $value
		727
		728	Forces the value to be encoded as (UTF-8) text values.
		729
		730	=item CBOR::XS::as_bytes $value
		731
		732	Forces the value to be encoded as a (binary) string value.
		733
		734	Example: encode a perl string as binary even though C<text_strings> is in
		735	effect.
		736
		737	CBOR::XS->new->text_strings->encode ([4, "text", CBOR::XS::bytes "bytevalue"]);
		738
		739	=item CBOR::XS::as_bool $value
		740
		741	Converts a Perl boolean (which can be any kind of scalar) into a CBOR
		742	boolean. Strictly the same, but shorter to write, than:
		743
		744	$value ? Types::Serialiser::true : Types::Serialiser::false
		745
		746	=item CBOR::XS::as_float16 $value
		747
		748	Forces half-float (IEEE 754 binary16) encoding of the given value.
		749
		750	=item CBOR::XS::as_float32 $value
		751
		752	Forces single-float (IEEE 754 binary32) encoding of the given value.
		753
		754	=item CBOR::XS::as_float64 $value
		755
		756	Forces double-float (IEEE 754 binary64) encoding of the given value.
		757
		758	=item CBOR::XS::as_cbor $cbor_text
		759
		760	Not a type cast per-se, this type cast forces the argument to eb encoded
		761	as-is. This can be used to embed pre-encoded CBOR data.
		762
		763	Note that no checking on the validity of the C<$cbor_text> is done - it's
		764	the callers responsibility to correctly encode values.
		765
		766	=item CBOR::XS::as_map [key => value...]
		767
		768	Treat the array reference as key value pairs and output a CBOR map. This
		769	allows you to generate CBOR maps with arbitrary key types (or, if you
		770	don't care about semantics, duplicate keys or prairs in a custom order),
		771	which is otherwise hard to do with Perl.
		772
		773	The single argument must be an array reference with an even number of
		774	elements.
		775
		776	Example: encode a CBOR map with a string and an integer as keys.
		777
		778	encode_cbor CBOR::XS::as_map [string => "value", 5 => "value"]
		779
		780	=back
		781
		782	=cut
		783
		784	sub CBOR::XS::as_cbor ($) { bless [$_[0], 0, undef], CBOR::XS::Tagged:: }
		785	sub CBOR::XS::as_int ($) { bless [$_[0], 1, undef], CBOR::XS::Tagged:: }
		786	sub CBOR::XS::as_bytes ($) { bless [$_[0], 2, undef], CBOR::XS::Tagged:: }
		787	sub CBOR::XS::as_text ($) { bless [$_[0], 3, undef], CBOR::XS::Tagged:: }
		788	sub CBOR::XS::as_float16 ($) { bless [$_[0], 4, undef], CBOR::XS::Tagged:: }
		789	sub CBOR::XS::as_float32 ($) { bless [$_[0], 5, undef], CBOR::XS::Tagged:: }
		790	sub CBOR::XS::as_float64 ($) { bless [$_[0], 6, undef], CBOR::XS::Tagged:: }
		791
		792	sub CBOR::XS::as_bool ($) { $_[0] ? $Types::Serialiser::true : $Types::Serialiser::false }
		793
		794	sub CBOR::XS::as_map ($) {
		795	ARRAY:: eq ref $_[0]
		796	and $#{ $_[0] } & 1
		797	or do { require Carp; Carp::croak ("CBOR::XS::as_map only acepts array references with an even number of elements, caught") };
		798
		799	bless [$_[0], 7, undef], CBOR::XS::Tagged::
		800	}
		801
		802	=head2 OBJECT SERIALISATION
		803
		804	This module implements both a CBOR-specific and the generic
		805	L<Types::Serialier> object serialisation protocol. The following
		806	subsections explain both methods.
		807
		808	=head3 ENCODING
		809
		810	This module knows two way to serialise a Perl object: The CBOR-specific
		811	way, and the generic way.
		812
		813	Whenever the encoder encounters a Perl object that it cannot serialise
		814	directly (most of them), it will first look up the C<TO_CBOR> method on
		815	it.
		816
		817	If it has a C<TO_CBOR> method, it will call it with the object as only
		818	argument, and expects exactly one return value, which it will then
		819	substitute and encode it in the place of the object.
		820
		821	Otherwise, it will look up the C<FREEZE> method. If it exists, it will
		822	call it with the object as first argument, and the constant string C<CBOR>
		823	as the second argument, to distinguish it from other serialisers.
		824
		825	The C<FREEZE> method can return any number of values (i.e. zero or
		826	more). These will be encoded as CBOR perl object, together with the
		827	classname.
		828
		829	These methods I<MUST NOT> change the data structure that is being
		830	serialised. Failure to comply to this can result in memory corruption -
		831	and worse.
		832
		833	If an object supports neither C<TO_CBOR> nor C<FREEZE>, encoding will fail
		834	with an error.
		835
		836	=head3 DECODING
		837
		838	Objects encoded via C<TO_CBOR> cannot (normally) be automatically decoded,
		839	but objects encoded via C<FREEZE> can be decoded using the following
		840	protocol:
		841
		842	When an encoded CBOR perl object is encountered by the decoder, it will
		843	look up the C<THAW> method, by using the stored classname, and will fail
		844	if the method cannot be found.
		845
		846	After the lookup it will call the C<THAW> method with the stored classname
		847	as first argument, the constant string C<CBOR> as second argument, and all
		848	values returned by C<FREEZE> as remaining arguments.
		849
		850	=head3 EXAMPLES
		851
		852	Here is an example C<TO_CBOR> method:
		853
		854	sub My::Object::TO_CBOR {
		855	my ($obj) = @_;
		856
		857	["this is a serialised My::Object object", $obj->{id}]
		858	}
		859
		860	When a C<My::Object> is encoded to CBOR, it will instead encode a simple
		861	array with two members: a string, and the "object id". Decoding this CBOR
		862	string will yield a normal perl array reference in place of the object.
		863
		864	A more useful and practical example would be a serialisation method for
		865	the URI module. CBOR has a custom tag value for URIs, namely 32:
		866
		867	sub URI::TO_CBOR {
		868	my ($self) = @_;
		869	my $uri = "$self"; # stringify uri
		870	utf8::upgrade $uri; # make sure it will be encoded as UTF-8 string
		871	CBOR::XS::tag 32, "$_[0]"
		872	}
		873
		874	This will encode URIs as a UTF-8 string with tag 32, which indicates an
		875	URI.
		876
		877	Decoding such an URI will not (currently) give you an URI object, but
		878	instead a CBOR::XS::Tagged object with tag number 32 and the string -
		879	exactly what was returned by C<TO_CBOR>.
		880
		881	To serialise an object so it can automatically be deserialised, you need
		882	to use C<FREEZE> and C<THAW>. To take the URI module as example, this
		883	would be a possible implementation:
		884
		885	sub URI::FREEZE {
		886	my ($self, $serialiser) = @_;
		887	"$self" # encode url string
		888	}
		889
		890	sub URI::THAW {
		891	my ($class, $serialiser, $uri) = @_;
		892	$class->new ($uri)
		893	}
		894
		895	Unlike C<TO_CBOR>, multiple values can be returned by C<FREEZE>. For
		896	example, a C<FREEZE> method that returns "type", "id" and "variant" values
		897	would cause an invocation of C<THAW> with 5 arguments:
		898
		899	sub My::Object::FREEZE {
		900	my ($self, $serialiser) = @_;
		901
		902	($self->{type}, $self->{id}, $self->{variant})
		903	}
		904
		905	sub My::Object::THAW {
		906	my ($class, $serialiser, $type, $id, $variant) = @_;
		907
		908	$class-<new (type => $type, id => $id, variant => $variant)
		909	}
		910
		911
		912	=head1 MAGIC HEADER
		913
		914	There is no way to distinguish CBOR from other formats
		915	programmatically. To make it easier to distinguish CBOR from other
		916	formats, the CBOR specification has a special "magic string" that can be
		917	prepended to any CBOR string without changing its meaning.
		918
		919	This string is available as C<$CBOR::XS::MAGIC>. This module does not
		920	prepend this string to the CBOR data it generates, but it will ignore it
		921	if present, so users can prepend this string as a "file type" indicator as
		922	required.
		923
		924
		925	=head1 THE CBOR::XS::Tagged CLASS
		926
		927	CBOR has the concept of tagged values - any CBOR value can be tagged with
		928	a numeric 64 bit number, which are centrally administered.
		929
		930	C<CBOR::XS> handles a few tags internally when en- or decoding. You can
		931	also create tags yourself by encoding C<CBOR::XS::Tagged> objects, and the
		932	decoder will create C<CBOR::XS::Tagged> objects itself when it hits an
		933	unknown tag.
		934
		935	These objects are simply blessed array references - the first member of
		936	the array being the numerical tag, the second being the value.
		937
		938	You can interact with C<CBOR::XS::Tagged> objects in the following ways:
		939
		940	=over 4
		941
		942	=item $tagged = CBOR::XS::tag $tag, $value
		943
		944	This function(!) creates a new C<CBOR::XS::Tagged> object using the given
		945	C<$tag> (0..2**64-1) to tag the given C<$value> (which can be any Perl
		946	value that can be encoded in CBOR, including serialisable Perl objects and
		947	C<CBOR::XS::Tagged> objects).
		948
		949	=item $tagged->[0]
		950
		951	=item $tagged->[0] = $new_tag
		952
		953	=item $tag = $tagged->tag
		954
		955	=item $new_tag = $tagged->tag ($new_tag)
		956
		957	Access/mutate the tag.
		958
		959	=item $tagged->[1]
		960
		961	=item $tagged->[1] = $new_value
		962
		963	=item $value = $tagged->value
		964
		965	=item $new_value = $tagged->value ($new_value)
		966
		967	Access/mutate the tagged value.
		968
		969	=back
		970
		971	=cut
		972
		973	sub tag($$) {
		974	bless [@_], CBOR::XS::Tagged::;
		975	}
		976
		977	sub CBOR::XS::Tagged::tag {
		978	$_[0][0] = $_[1] if $#_;
		979	$_[0][0]
		980	}
		981
		982	sub CBOR::XS::Tagged::value {
		983	$_[0][1] = $_[1] if $#_;
		984	$_[0][1]
		985	}
		986
		987	=head2 EXAMPLES
		988
		989	Here are some examples of C<CBOR::XS::Tagged> uses to tag objects.
		990
		991	You can look up CBOR tag value and emanings in the IANA registry at
		992	L<http://www.iana.org/assignments/cbor-tags/cbor-tags.xhtml>.
		993
		994	Prepend a magic header (C<$CBOR::XS::MAGIC>):
		995
		996	my $cbor = encode_cbor CBOR::XS::tag 55799, $value;
		997	# same as:
		998	my $cbor = $CBOR::XS::MAGIC . encode_cbor $value;
		999
		1000	Serialise some URIs and a regex in an array:
		1001
		1002	my $cbor = encode_cbor [
		1003	(CBOR::XS::tag 32, "http://www.nethype.de/"),
		1004	(CBOR::XS::tag 32, "http://software.schmorp.de/"),
		1005	(CBOR::XS::tag 35, "^[Pp][Ee][Rr][lL]\$"),
		1006	];
		1007
		1008	Wrap CBOR data in CBOR:
		1009
		1010	my $cbor_cbor = encode_cbor
		1011	CBOR::XS::tag 24,
		1012	encode_cbor [1, 2, 3];
		1013
		1014	=head1 TAG HANDLING AND EXTENSIONS
		1015
		1016	This section describes how this module handles specific tagged values
		1017	and extensions. If a tag is not mentioned here and no additional filters
		1018	are provided for it, then the default handling applies (creating a
		1019	CBOR::XS::Tagged object on decoding, and only encoding the tag when
		1020	explicitly requested).
		1021
		1022	Tags not handled specifically are currently converted into a
		1023	L<CBOR::XS::Tagged> object, which is simply a blessed array reference
		1024	consisting of the numeric tag value followed by the (decoded) CBOR value.
		1025
		1026	Future versions of this module reserve the right to special case
		1027	additional tags (such as base64url).
		1028
		1029	=head2 ENFORCED TAGS
		1030
		1031	These tags are always handled when decoding, and their handling cannot be
		1032	overridden by the user.
		1033
		1034	=over 4
		1035
		1036	=item 26 (perl-object, L<http://cbor.schmorp.de/perl-object>)
		1037
		1038	These tags are automatically created (and decoded) for serialisable
		1039	objects using the C<FREEZE/THAW> methods (the L<Types::Serialier> object
		1040	serialisation protocol). See L<OBJECT SERIALISATION> for details.
		1041
		1042	=item 28, 29 (shareable, sharedref, L<http://cbor.schmorp.de/value-sharing>)
		1043
		1044	These tags are automatically decoded when encountered (and they do not
		1045	result in a cyclic data structure, see C<allow_cycles>), resulting in
		1046	shared values in the decoded object. They are only encoded, however, when
		1047	C<allow_sharing> is enabled.
		1048
		1049	Not all shared values can be successfully decoded: values that reference
		1050	themselves will I<currently> decode as C<undef> (this is not the same
		1051	as a reference pointing to itself, which will be represented as a value
		1052	that contains an indirect reference to itself - these will be decoded
		1053	properly).
		1054
		1055	Note that considerably more shared value data structures can be decoded
		1056	than will be encoded - currently, only values pointed to by references
		1057	will be shared, others will not. While non-reference shared values can be
		1058	generated in Perl with some effort, they were considered too unimportant
		1059	to be supported in the encoder. The decoder, however, will decode these
		1060	values as shared values.
		1061
		1062	=item 256, 25 (stringref-namespace, stringref, L<http://cbor.schmorp.de/stringref>)
		1063
		1064	These tags are automatically decoded when encountered. They are only
		1065	encoded, however, when C<pack_strings> is enabled.
		1066
		1067	=item 22098 (indirection, L<http://cbor.schmorp.de/indirection>)
		1068
		1069	This tag is automatically generated when a reference are encountered (with
		1070	the exception of hash and array references). It is converted to a reference
		1071	when decoding.
		1072
		1073	=item 55799 (self-describe CBOR, RFC 7049)
		1074
		1075	This value is not generated on encoding (unless explicitly requested by
		1076	the user), and is simply ignored when decoding.
		1077
		1078	=back
		1079
		1080	=head2 NON-ENFORCED TAGS
		1081
		1082	These tags have default filters provided when decoding. Their handling can
		1083	be overridden by changing the C<%CBOR::XS::FILTER> entry for the tag, or by
		1084	providing a custom C<filter> callback when decoding.
		1085
		1086	When they result in decoding into a specific Perl class, the module
		1087	usually provides a corresponding C<TO_CBOR> method as well.
		1088
		1089	When any of these need to load additional modules that are not part of the
		1090	perl core distribution (e.g. L<URI>), it is (currently) up to the user to
		1091	provide these modules. The decoding usually fails with an exception if the
		1092	required module cannot be loaded.
		1093
		1094	=over 4
		1095
		1096	=item 0, 1 (date/time string, seconds since the epoch)
		1097
		1098	These tags are decoded into L<Time::Piece> objects. The corresponding
		1099	C<Time::Piece::TO_CBOR> method always encodes into tag 1 values currently.
		1100
		1101	The L<Time::Piece> API is generally surprisingly bad, and fractional
		1102	seconds are only accidentally kept intact, so watch out. On the plus side,
		1103	the module comes with perl since 5.10, which has to count for something.
		1104
		1105	=item 2, 3 (positive/negative bignum)
		1106
		1107	These tags are decoded into L<Math::BigInt> objects. The corresponding
		1108	C<Math::BigInt::TO_CBOR> method encodes "small" bigints into normal CBOR
		1109	integers, and others into positive/negative CBOR bignums.
		1110
		1111	=item 4, 5, 264, 265 (decimal fraction/bigfloat)
		1112
		1113	Both decimal fractions and bigfloats are decoded into L<Math::BigFloat>
		1114	objects. The corresponding C<Math::BigFloat::TO_CBOR> method I<always>
		1115	encodes into a decimal fraction (either tag 4 or 264).
		1116
		1117	NaN and infinities are not encoded properly, as they cannot be represented
		1118	in CBOR.
		1119
		1120	See L<BIGNUM SECURITY CONSIDERATIONS> for more info.
		1121
		1122	=item 30 (rational numbers)
		1123
		1124	These tags are decoded into L<Math::BigRat> objects. The corresponding
		1125	C<Math::BigRat::TO_CBOR> method encodes rational numbers with denominator
		1126	C<1> via their numerator only, i.e., they become normal integers or
		1127	C<bignums>.
		1128
		1129	See L<BIGNUM SECURITY CONSIDERATIONS> for more info.
		1130
		1131	=item 21, 22, 23 (expected later JSON conversion)
		1132
		1133	CBOR::XS is not a CBOR-to-JSON converter, and will simply ignore these
		1134	tags.
		1135
		1136	=item 32 (URI)
		1137
		1138	These objects decode into L<URI> objects. The corresponding
		1139	C<URI::TO_CBOR> method again results in a CBOR URI value.
		1140
		1141	=back
		1142
		1143	=cut
		1144
265	=head2 CBOR and JSON	1145	=head1 CBOR and JSON
266		1146
267	TODO	1147	CBOR is supposed to implement a superset of the JSON data model, and is,
		1148	with some coercion, able to represent all JSON texts (something that other
		1149	"binary JSON" formats such as BSON generally do not support).
		1150
		1151	CBOR implements some extra hints and support for JSON interoperability,
		1152	and the spec offers further guidance for conversion between CBOR and
		1153	JSON. None of this is currently implemented in CBOR, and the guidelines
		1154	in the spec do not result in correct round-tripping of data. If JSON
		1155	interoperability is improved in the future, then the goal will be to
		1156	ensure that decoded JSON data will round-trip encoding and decoding to
		1157	CBOR intact.
268		1158
269		1159
270	=head1 SECURITY CONSIDERATIONS	1160	=head1 SECURITY CONSIDERATIONS
271		1161
272	When you are using CBOR in a protocol, talking to untrusted potentially	1162	Tl;dr... if you want to decode or encode CBOR from untrusted sources, you
273	hostile creatures requires relatively few measures.	1163	should start with a coder object created via C<new_safe> (which implements
		1164	the mitigations explained below):
274		1165
		1166	my $coder = CBOR::XS->new_safe;
		1167
		1168	my $data = $coder->decode ($cbor_text);
		1169	my $cbor = $coder->encode ($data);
		1170
		1171	Longer version: When you are using CBOR in a protocol, talking to
		1172	untrusted potentially hostile creatures requires some thought:
		1173
		1174	=over 4
		1175
		1176	=item Security of the CBOR decoder itself
		1177
275	First of all, your CBOR decoder should be secure, that is, should not have	1178	First and foremost, your CBOR decoder should be secure, that is, should
		1179	not have any buffer overflows or similar bugs that could potentially be
276	any buffer overflows. Obviously, this module should ensure that and I am	1180	exploited. Obviously, this module should ensure that and I am trying hard
277	trying hard on making that true, but you never know.	1181	on making that true, but you never know.
278		1182
		1183	=item CBOR::XS can invoke almost arbitrary callbacks during decoding
		1184
		1185	CBOR::XS supports object serialisation - decoding CBOR can cause calls
		1186	to I<any> C<THAW> method in I<any> package that exists in your process
		1187	(that is, CBOR::XS will not try to load modules, but any existing C<THAW>
		1188	method or function can be called, so they all have to be secure).
		1189
		1190	Less obviously, it will also invoke C<TO_CBOR> and C<FREEZE> methods -
		1191	even if all your C<THAW> methods are secure, encoding data structures from
		1192	untrusted sources can invoke those and trigger bugs in those.
		1193
		1194	So, if you are not sure about the security of all the modules you
		1195	have loaded (you shouldn't), you should disable this part using
		1196	C<forbid_objects> or using C<new_safe>.
		1197
		1198	=item CBOR can be extended with tags that call library code
		1199
		1200	CBOR can be extended with tags, and C<CBOR::XS> has a registry of
		1201	conversion functions for many existing tags that can be extended via
		1202	third-party modules (see the C<filter> method).
		1203
		1204	If you don't trust these, you should configure the "safe" filter function,
		1205	C<CBOR::XS::safe_filter> (C<new_safe> does this), which by default only
		1206	includes conversion functions that are considered "safe" by the author
		1207	(but again, they can be extended by third party modules).
		1208
		1209	Depending on your level of paranoia, you can use the "safe" filter:
		1210
		1211	$cbor->filter (\&CBOR::XS::safe_filter);
		1212
		1213	... your own filter...
		1214
		1215	$cbor->filter (sub { ... do your stuffs here ... });
		1216
		1217	... or even no filter at all, disabling all tag decoding:
		1218
		1219	$cbor->filter (sub { });
		1220
		1221	This is never a problem for encoding, as the tag mechanism only exists in
		1222	CBOR texts.
		1223
		1224	=item Resource-starving attacks: object memory usage
		1225
279	Second, you need to avoid resource-starving attacks. That means you should	1226	You need to avoid resource-starving attacks. That means you should limit
280	limit the size of CBOR data you accept, or make sure then when your	1227	the size of CBOR data you accept, or make sure then when your resources
281	resources run out, that's just fine (e.g. by using a separate process that	1228	run out, that's just fine (e.g. by using a separate process that can
282	can crash safely). The size of a CBOR string in octets is usually a good	1229	crash safely). The size of a CBOR string in octets is usually a good
283	indication of the size of the resources required to decode it into a Perl	1230	indication of the size of the resources required to decode it into a Perl
284	structure. While CBOR::XS can check the size of the CBOR text, it might be	1231	structure. While CBOR::XS can check the size of the CBOR text (using
285	too late when you already have it in memory, so you might want to check	1232	C<max_size> - done by C<new_safe>), it might be too late when you already
286	the size before you accept the string.	1233	have it in memory, so you might want to check the size before you accept
		1234	the string.
287		1235
		1236	As for encoding, it is possible to construct data structures that are
		1237	relatively small but result in large CBOR texts (for example by having an
		1238	array full of references to the same big data structure, which will all be
		1239	deep-cloned during encoding by default). This is rarely an actual issue
		1240	(and the worst case is still just running out of memory), but you can
		1241	reduce this risk by using C<allow_sharing>.
		1242
		1243	=item Resource-starving attacks: stack overflows
		1244
288	Third, CBOR::XS recurses using the C stack when decoding objects and	1245	CBOR::XS recurses using the C stack when decoding objects and arrays. The
289	arrays. The C stack is a limited resource: for instance, on my amd64	1246	C stack is a limited resource: for instance, on my amd64 machine with 8MB
290	machine with 8MB of stack size I can decode around 180k nested arrays but	1247	of stack size I can decode around 180k nested arrays but only 14k nested
291	only 14k nested CBOR objects (due to perl itself recursing deeply on croak	1248	CBOR objects (due to perl itself recursing deeply on croak to free the
292	to free the temporary). If that is exceeded, the program crashes. To be	1249	temporary). If that is exceeded, the program crashes. To be conservative,
293	conservative, the default nesting limit is set to 512. If your process	1250	the default nesting limit is set to 512. If your process has a smaller
294	has a smaller stack, you should adjust this setting accordingly with the	1251	stack, you should adjust this setting accordingly with the C<max_depth>
295	C<max_depth> method.	1252	method.
		1253
		1254	=item Resource-starving attacks: CPU en-/decoding complexity
		1255
		1256	CBOR::XS will use the L<Math::BigInt>, L<Math::BigFloat> and
		1257	L<Math::BigRat> libraries to represent encode/decode bignums. These can be
		1258	very slow (as in, centuries of CPU time) and can even crash your program
		1259	(and are generally not very trustworthy). See the next section on bignum
		1260	security for details.
		1261
		1262	=item Data breaches: leaking information in error messages
		1263
		1264	CBOR::XS might leak contents of your Perl data structures in its error
		1265	messages, so when you serialise sensitive information you might want to
		1266	make sure that exceptions thrown by CBOR::XS will not end up in front of
		1267	untrusted eyes.
		1268
		1269	=item Something else...
296		1270
297	Something else could bomb you, too, that I forgot to think of. In that	1271	Something else could bomb you, too, that I forgot to think of. In that
298	case, you get to keep the pieces. I am always open for hints, though...	1272	case, you get to keep the pieces. I am always open for hints, though...
299		1273
300	Also keep in mind that CBOR::XS might leak contents of your Perl data	1274	=back
301	structures in its error messages, so when you serialise sensitive	1275
302	information you might want to make sure that exceptions thrown by CBOR::XS	1276
303	will not end up in front of untrusted eyes.	1277	=head1 BIGNUM SECURITY CONSIDERATIONS
		1278
		1279	CBOR::XS provides a C<TO_CBOR> method for both L<Math::BigInt> and
		1280	L<Math::BigFloat> that tries to encode the number in the simplest possible
		1281	way, that is, either a CBOR integer, a CBOR bigint/decimal fraction (tag
		1282	4) or an arbitrary-exponent decimal fraction (tag 264). Rational numbers
		1283	(L<Math::BigRat>, tag 30) can also contain bignums as members.
		1284
		1285	CBOR::XS will also understand base-2 bigfloat or arbitrary-exponent
		1286	bigfloats (tags 5 and 265), but it will never generate these on its own.
		1287
		1288	Using the built-in L<Math::BigInt::Calc> support, encoding and decoding
		1289	decimal fractions is generally fast. Decoding bigints can be slow for very
		1290	big numbers (tens of thousands of digits, something that could potentially
		1291	be caught by limiting the size of CBOR texts), and decoding bigfloats or
		1292	arbitrary-exponent bigfloats can be I<extremely> slow (minutes, decades)
		1293	for large exponents (roughly 40 bit and longer).
		1294
		1295	Additionally, L<Math::BigInt> can take advantage of other bignum
		1296	libraries, such as L<Math::GMP>, which cannot handle big floats with large
		1297	exponents, and might simply abort or crash your program, due to their code
		1298	quality.
		1299
		1300	This can be a concern if you want to parse untrusted CBOR. If it is, you
		1301	might want to disable decoding of tag 2 (bigint) and 3 (negative bigint)
		1302	types. You should also disable types 5 and 265, as these can be slow even
		1303	without bigints.
		1304
		1305	Disabling bigints will also partially or fully disable types that rely on
		1306	them, e.g. rational numbers that use bignums.
		1307
304		1308
305	=head1 CBOR IMPLEMENTATION NOTES	1309	=head1 CBOR IMPLEMENTATION NOTES
306		1310
307	This section contains some random implementation notes. They do not	1311	This section contains some random implementation notes. They do not
308	describe guaranteed behaviour, but merely behaviour as-is implemented	1312	describe guaranteed behaviour, but merely behaviour as-is implemented
…		…
317	Only the double data type is supported for NV data types - when Perl uses	1321	Only the double data type is supported for NV data types - when Perl uses
318	long double to represent floating point values, they might not be encoded	1322	long double to represent floating point values, they might not be encoded
319	properly. Half precision types are accepted, but not encoded.	1323	properly. Half precision types are accepted, but not encoded.
320		1324
321	Strict mode and canonical mode are not implemented.	1325	Strict mode and canonical mode are not implemented.
		1326
		1327
		1328	=head1 LIMITATIONS ON PERLS WITHOUT 64-BIT INTEGER SUPPORT
		1329
		1330	On perls that were built without 64 bit integer support (these are rare
		1331	nowadays, even on 32 bit architectures, as all major Perl distributions
		1332	are built with 64 bit integer support), support for any kind of 64 bit
		1333	value in CBOR is very limited - most likely, these 64 bit values will
		1334	be truncated, corrupted, or otherwise not decoded correctly. This also
		1335	includes string, float, array and map sizes that are stored as 64 bit
		1336	integers.
322		1337
323		1338
324	=head1 THREADS	1339	=head1 THREADS
325		1340
326	This module is I<not> guaranteed to be thread safe and there are no	1341	This module is I<not> guaranteed to be thread safe and there are no
…		…
340	Please refrain from using rt.cpan.org or any other bug reporting	1355	Please refrain from using rt.cpan.org or any other bug reporting
341	service. I put the contact address into my modules for a reason.	1356	service. I put the contact address into my modules for a reason.
342		1357
343	=cut	1358	=cut
344		1359
345	our $true = do { bless \(my $dummy = 1), "CBOR::XS::Boolean" };	1360	# clumsy and slow hv_store-in-hash helper function
346	our $false = do { bless \(my $dummy = 0), "CBOR::XS::Boolean" };	1361	sub _hv_store {
347		1362	$_[0]{$_[1]} = $_[2];
348	sub true() { $true }
349	sub false() { $false }
350
351	sub is_bool($) {
352	UNIVERSAL::isa $_[0], "CBOR::XS::Boolean"
353	# or UNIVERSAL::isa $_[0], "CBOR::Literal"
354	}	1363	}
355		1364
		1365	our %FILTER = (
		1366	0 => sub { # rfc4287 datetime, utf-8
		1367	require Time::Piece;
		1368	# Time::Piece::Strptime uses the "incredibly flexible date parsing routine"
		1369	# from FreeBSD, which can't parse ISO 8601, RFC3339, RFC4287 or much of anything
		1370	# else either. Whats incredibe over standard strptime totally escapes me.
		1371	# doesn't do fractional times, either. sigh.
		1372	# In fact, it's all a lie, it uses whatever strptime it wants, and of course,
		1373	# they are all incompatible. The openbsd one simply ignores %z (but according to the
		1374	# docs, it would be much more incredibly flexible indeed. If it worked, that is.).
		1375	scalar eval {
		1376	my $s = $_[1];
		1377
		1378	$s =~ s/Z$/+00:00/;
		1379	$s =~ s/(\.[0-9]+)?([+-][0-9][0-9]):([0-9][0-9])$//
		1380	or die;
		1381
		1382	my $b = $1 - ($2 * 60 + $3) * 60; # fractional part + offset. hopefully
		1383	my $d = Time::Piece->strptime ($s, "%Y-%m-%dT%H:%M:%S");
		1384
		1385	Time::Piece::gmtime ($d->epoch + $b)
		1386	} || die "corrupted CBOR date/time string ($_[0])";
		1387	},
		1388
		1389	1 => sub { # seconds since the epoch, possibly fractional
		1390	require Time::Piece;
		1391	scalar Time::Piece::gmtime (pop)
		1392	},
		1393
		1394	2 => sub { # pos bigint
		1395	require Math::BigInt;
		1396	Math::BigInt->new ("0x" . unpack "H*", pop)
		1397	},
		1398
		1399	3 => sub { # neg bigint
		1400	require Math::BigInt;
		1401	-Math::BigInt->new ("0x" . unpack "H*", pop)
		1402	},
		1403
		1404	4 => sub { # decimal fraction, array
		1405	require Math::BigFloat;
		1406	Math::BigFloat->new ($_[1][1] . "E" . $_[1][0])
		1407	},
		1408
		1409	264 => sub { # decimal fraction with arbitrary exponent
		1410	require Math::BigFloat;
		1411	Math::BigFloat->new ($_[1][1] . "E" . $_[1][0])
		1412	},
		1413
		1414	5 => sub { # bigfloat, array
		1415	require Math::BigFloat;
		1416	scalar Math::BigFloat->new ($_[1][1]) * Math::BigFloat->new (2)->bpow ($_[1][0])
		1417	},
		1418
		1419	265 => sub { # bigfloat with arbitrary exponent
		1420	require Math::BigFloat;
		1421	scalar Math::BigFloat->new ($_[1][1]) * Math::BigFloat->new (2)->bpow ($_[1][0])
		1422	},
		1423
		1424	30 => sub { # rational number
		1425	require Math::BigRat;
		1426	Math::BigRat->new ("$_[1][0]/$_[1][1]") # separate parameters only work in recent versons
		1427	},
		1428
		1429	21 => sub { pop }, # expected conversion to base64url encoding
		1430	22 => sub { pop }, # expected conversion to base64 encoding
		1431	23 => sub { pop }, # expected conversion to base16 encoding
		1432
		1433	# 24 # embedded cbor, byte string
		1434
		1435	32 => sub {
		1436	require URI;
		1437	URI->new (pop)
		1438	},
		1439
		1440	# 33 # base64url rfc4648, utf-8
		1441	# 34 # base64 rfc46484, utf-8
		1442	# 35 # regex pcre/ecma262, utf-8
		1443	# 36 # mime message rfc2045, utf-8
		1444	);
		1445
		1446	sub default_filter {
		1447	&{ $FILTER{$_[0]} or return }
		1448	}
		1449
		1450	our %SAFE_FILTER = map { $_ => $FILTER{$_} } 0, 1, 21, 22, 23, 32;
		1451
		1452	sub safe_filter {
		1453	&{ $SAFE_FILTER{$_[0]} or return }
		1454	}
		1455
		1456	sub URI::TO_CBOR {
		1457	my $uri = $_[0]->as_string;
		1458	utf8::upgrade $uri;
		1459	tag 32, $uri
		1460	}
		1461
		1462	sub Math::BigInt::TO_CBOR {
		1463	if (-2147483648 <= $_[0] && $_[0] <= 2147483647) {
		1464	$_[0]->numify
		1465	} else {
		1466	my $hex = substr $_[0]->as_hex, 2;
		1467	$hex = "0$hex" if 1 & length $hex; # sigh
		1468	tag $_[0] >= 0 ? 2 : 3, pack "H*", $hex
		1469	}
		1470	}
		1471
		1472	sub Math::BigFloat::TO_CBOR {
		1473	my ($m, $e) = $_[0]->parts;
		1474
		1475	-9223372036854775808 <= $e && $e <= 18446744073709551615
		1476	? tag 4, [$e->numify, $m]
		1477	: tag 264, [$e, $m]
		1478	}
		1479
		1480	sub Math::BigRat::TO_CBOR {
		1481	my ($n, $d) = $_[0]->parts;
		1482
		1483	# older versions of BigRat need *1, as they not always return numbers
		1484
		1485	$d*1 == 1
		1486	? $n*1
		1487	: tag 30, [$n*1, $d*1]
		1488	}
		1489
		1490	sub Time::Piece::TO_CBOR {
		1491	tag 1, 0 + $_[0]->epoch
		1492	}
		1493
356	XSLoader::load "CBOR::XS", $VERSION;	1494	XSLoader::load "CBOR::XS", $VERSION;
357
358	package CBOR::XS::Boolean;
359
360	use overload
361	"0+" => sub { ${$_[0]} },
362	"++" => sub { $_[0] = ${$_[0]} + 1 },
363	"--" => sub { $_[0] = ${$_[0]} - 1 },
364	fallback => 1;
365
366	1;
367		1495
368	=head1 SEE ALSO	1496	=head1 SEE ALSO
369		1497
370	The L<JSON> and L<JSON::XS> modules that do similar, but human-readable,	1498	The L<JSON> and L<JSON::XS> modules that do similar, but human-readable,
371	serialisation.	1499	serialisation.
372		1500
		1501	The L<Types::Serialiser> module provides the data model for true, false
		1502	and error values.
		1503
373	=head1 AUTHOR	1504	=head1 AUTHOR
374		1505
375	Marc Lehmann <schmorp@schmorp.de>	1506	Marc Lehmann <schmorp@schmorp.de>
376	http://home.schmorp.de/	1507	http://home.schmorp.de/
377		1508
378	=cut	1509	=cut
379		1510
		1511	1
		1512

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines