1 |
NAME |
2 |
CBOR::XS - Concise Binary Object Representation (CBOR, RFC7049) |
3 |
|
4 |
SYNOPSIS |
5 |
use CBOR::XS; |
6 |
|
7 |
$binary_cbor_data = encode_cbor $perl_value; |
8 |
$perl_value = decode_cbor $binary_cbor_data; |
9 |
|
10 |
# OO-interface |
11 |
|
12 |
$coder = CBOR::XS->new; |
13 |
$binary_cbor_data = $coder->encode ($perl_value); |
14 |
$perl_value = $coder->decode ($binary_cbor_data); |
15 |
|
16 |
# prefix decoding |
17 |
|
18 |
my $many_cbor_strings = ...; |
19 |
while (length $many_cbor_strings) { |
20 |
my ($data, $length) = $cbor->decode_prefix ($many_cbor_strings); |
21 |
# data was decoded |
22 |
substr $many_cbor_strings, 0, $length, ""; # remove decoded cbor string |
23 |
} |
24 |
|
25 |
DESCRIPTION |
26 |
This module converts Perl data structures to the Concise Binary Object |
27 |
Representation (CBOR) and vice versa. CBOR is a fast binary |
28 |
serialisation format that aims to use an (almost) superset of the JSON |
29 |
data model, i.e. when you can represent something useful in JSON, you |
30 |
should be able to represent it in CBOR. |
31 |
|
32 |
In short, CBOR is a faster and quite compact binary alternative to JSON, |
33 |
with the added ability of supporting serialisation of Perl objects. |
34 |
(JSON often compresses better than CBOR though, so if you plan to |
35 |
compress the data later and speed is less important you might want to |
36 |
compare both formats first). |
37 |
|
38 |
To give you a general idea about speed, with texts in the megabyte |
39 |
range, "CBOR::XS" usually encodes roughly twice as fast as Storable or |
40 |
JSON::XS and decodes about 15%-30% faster than those. The shorter the |
41 |
data, the worse Storable performs in comparison. |
42 |
|
43 |
Regarding compactness, "CBOR::XS"-encoded data structures are usually |
44 |
about 20% smaller than the same data encoded as (compact) JSON or |
45 |
Storable. |
46 |
|
47 |
In addition to the core CBOR data format, this module implements a |
48 |
number of extensions, to support cyclic and shared data structures (see |
49 |
"allow_sharing" and "allow_cycles"), string deduplication (see |
50 |
"pack_strings") and scalar references (always enabled). |
51 |
|
52 |
The primary goal of this module is to be *correct* and the secondary |
53 |
goal is to be *fast*. To reach the latter goal it was written in C. |
54 |
|
55 |
See MAPPING, below, on how CBOR::XS maps perl values to CBOR values and |
56 |
vice versa. |
57 |
|
58 |
FUNCTIONAL INTERFACE |
59 |
The following convenience methods are provided by this module. They are |
60 |
exported by default: |
61 |
|
62 |
$cbor_data = encode_cbor $perl_scalar |
63 |
Converts the given Perl data structure to CBOR representation. |
64 |
Croaks on error. |
65 |
|
66 |
$perl_scalar = decode_cbor $cbor_data |
67 |
The opposite of "encode_cbor": expects a valid CBOR string to parse, |
68 |
returning the resulting perl scalar. Croaks on error. |
69 |
|
70 |
OBJECT-ORIENTED INTERFACE |
71 |
The object oriented interface lets you configure your own encoding or |
72 |
decoding style, within the limits of supported formats. |
73 |
|
74 |
$cbor = new CBOR::XS |
75 |
Creates a new CBOR::XS object that can be used to de/encode CBOR |
76 |
strings. All boolean flags described below are by default |
77 |
*disabled*. |
78 |
|
79 |
The mutators for flags all return the CBOR object again and thus |
80 |
calls can be chained: |
81 |
|
82 |
my $cbor = CBOR::XS->new->encode ({a => [1,2]}); |
83 |
|
84 |
$cbor = $cbor->max_depth ([$maximum_nesting_depth]) |
85 |
$max_depth = $cbor->get_max_depth |
86 |
Sets the maximum nesting level (default 512) accepted while encoding |
87 |
or decoding. If a higher nesting level is detected in CBOR data or a |
88 |
Perl data structure, then the encoder and decoder will stop and |
89 |
croak at that point. |
90 |
|
91 |
Nesting level is defined by number of hash- or arrayrefs that the |
92 |
encoder needs to traverse to reach a given point or the number of |
93 |
"{" or "[" characters without their matching closing parenthesis |
94 |
crossed to reach a given character in a string. |
95 |
|
96 |
Setting the maximum depth to one disallows any nesting, so that |
97 |
ensures that the object is only a single hash/object or array. |
98 |
|
99 |
If no argument is given, the highest possible setting will be used, |
100 |
which is rarely useful. |
101 |
|
102 |
Note that nesting is implemented by recursion in C. The default |
103 |
value has been chosen to be as large as typical operating systems |
104 |
allow without crashing. |
105 |
|
106 |
See SECURITY CONSIDERATIONS, below, for more info on why this is |
107 |
useful. |
108 |
|
109 |
$cbor = $cbor->max_size ([$maximum_string_size]) |
110 |
$max_size = $cbor->get_max_size |
111 |
Set the maximum length a CBOR string may have (in bytes) where |
112 |
decoding is being attempted. The default is 0, meaning no limit. |
113 |
When "decode" is called on a string that is longer then this many |
114 |
bytes, it will not attempt to decode the string but throw an |
115 |
exception. This setting has no effect on "encode" (yet). |
116 |
|
117 |
If no argument is given, the limit check will be deactivated (same |
118 |
as when 0 is specified). |
119 |
|
120 |
See SECURITY CONSIDERATIONS, below, for more info on why this is |
121 |
useful. |
122 |
|
123 |
$cbor = $cbor->allow_unknown ([$enable]) |
124 |
$enabled = $cbor->get_allow_unknown |
125 |
If $enable is true (or missing), then "encode" will *not* throw an |
126 |
exception when it encounters values it cannot represent in CBOR (for |
127 |
example, filehandles) but instead will encode a CBOR "error" value. |
128 |
|
129 |
If $enable is false (the default), then "encode" will throw an |
130 |
exception when it encounters anything it cannot encode as CBOR. |
131 |
|
132 |
This option does not affect "decode" in any way, and it is |
133 |
recommended to leave it off unless you know your communications |
134 |
partner. |
135 |
|
136 |
$cbor = $cbor->allow_sharing ([$enable]) |
137 |
$enabled = $cbor->get_allow_sharing |
138 |
If $enable is true (or missing), then "encode" will not |
139 |
double-encode values that have been referenced before (e.g. when the |
140 |
same object, such as an array, is referenced multiple times), but |
141 |
instead will emit a reference to the earlier value. |
142 |
|
143 |
This means that such values will only be encoded once, and will not |
144 |
result in a deep cloning of the value on decode, in decoders |
145 |
supporting the value sharing extension. This also makes it possible |
146 |
to encode cyclic data structures (which need "allow_cycles" to ne |
147 |
enabled to be decoded by this module). |
148 |
|
149 |
It is recommended to leave it off unless you know your communication |
150 |
partner supports the value sharing extensions to CBOR |
151 |
(<http://cbor.schmorp.de/value-sharing>), as without decoder |
152 |
support, the resulting data structure might be unusable. |
153 |
|
154 |
Detecting shared values incurs a runtime overhead when values are |
155 |
encoded that have a reference counter large than one, and might |
156 |
unnecessarily increase the encoded size, as potentially shared |
157 |
values are encode as shareable whether or not they are actually |
158 |
shared. |
159 |
|
160 |
At the moment, only targets of references can be shared (e.g. |
161 |
scalars, arrays or hashes pointed to by a reference). Weirder |
162 |
constructs, such as an array with multiple "copies" of the *same* |
163 |
string, which are hard but not impossible to create in Perl, are not |
164 |
supported (this is the same as with Storable). |
165 |
|
166 |
If $enable is false (the default), then "encode" will encode shared |
167 |
data structures repeatedly, unsharing them in the process. Cyclic |
168 |
data structures cannot be encoded in this mode. |
169 |
|
170 |
This option does not affect "decode" in any way - shared values and |
171 |
references will always be decoded properly if present. |
172 |
|
173 |
$cbor = $cbor->allow_cycles ([$enable]) |
174 |
$enabled = $cbor->get_allow_cycles |
175 |
If $enable is true (or missing), then "decode" will happily decode |
176 |
self-referential (cyclic) data structures. By default these will not |
177 |
be decoded, as they need manual cleanup to avoid memory leaks, so |
178 |
code that isn't prepared for this will not leak memory. |
179 |
|
180 |
If $enable is false (the default), then "decode" will throw an error |
181 |
when it encounters a self-referential/cyclic data structure. |
182 |
|
183 |
FUTURE DIRECTION: the motivation behind this option is to avoid |
184 |
*real* cycles - future versions of this module might chose to decode |
185 |
cyclic data structures using weak references when this option is |
186 |
off, instead of throwing an error. |
187 |
|
188 |
This option does not affect "encode" in any way - shared values and |
189 |
references will always be encoded properly if present. |
190 |
|
191 |
$cbor = $cbor->pack_strings ([$enable]) |
192 |
$enabled = $cbor->get_pack_strings |
193 |
If $enable is true (or missing), then "encode" will try not to |
194 |
encode the same string twice, but will instead encode a reference to |
195 |
the string instead. Depending on your data format, this can save a |
196 |
lot of space, but also results in a very large runtime overhead |
197 |
(expect encoding times to be 2-4 times as high as without). |
198 |
|
199 |
It is recommended to leave it off unless you know your |
200 |
communications partner supports the stringref extension to CBOR |
201 |
(<http://cbor.schmorp.de/stringref>), as without decoder support, |
202 |
the resulting data structure might not be usable. |
203 |
|
204 |
If $enable is false (the default), then "encode" will encode strings |
205 |
the standard CBOR way. |
206 |
|
207 |
This option does not affect "decode" in any way - string references |
208 |
will always be decoded properly if present. |
209 |
|
210 |
$cbor = $cbor->text_keys ([$enable]) |
211 |
$enabled = $cbor->get_text_keys |
212 |
If $enabled is true (or missing), then "encode" will encode all perl |
213 |
hash keys as CBOR text strings/UTF-8 string, upgrading them as |
214 |
needed. |
215 |
|
216 |
If $enable is false (the default), then "encode" will encode hash |
217 |
keys normally - upgraded perl strings (strings internally encoded as |
218 |
UTF-8) as CBOR text strings, and downgraded perl strings as CBOR |
219 |
byte strings. |
220 |
|
221 |
This option does not affect "decode" in any way. |
222 |
|
223 |
This option is useful for interoperability with CBOR decoders that |
224 |
don't treat byte strings as a form of text. It is especially useful |
225 |
as Perl gives very little control over hash keys. |
226 |
|
227 |
Enabling this option can be slow, as all downgraded hash keys that |
228 |
are encoded need to be scanned and converted to UTF-8. |
229 |
|
230 |
$cbor = $cbor->text_strings ([$enable]) |
231 |
$enabled = $cbor->get_text_strings |
232 |
This option works similar to "text_keys", above, but works on all |
233 |
strings (including hash keys), so "text_keys" has no further effect |
234 |
after enabling "text_strings". |
235 |
|
236 |
If $enabled is true (or missing), then "encode" will encode all perl |
237 |
strings as CBOR text strings/UTF-8 strings, upgrading them as |
238 |
needed. |
239 |
|
240 |
If $enable is false (the default), then "encode" will encode strings |
241 |
normally (but see "text_keys") - upgraded perl strings (strings |
242 |
internally encoded as UTF-8) as CBOR text strings, and downgraded |
243 |
perl strings as CBOR byte strings. |
244 |
|
245 |
This option does not affect "decode" in any way. |
246 |
|
247 |
This option has similar advantages and disadvantages as "text_keys". |
248 |
In addition, this option effectively removes the ability to encode |
249 |
byte strings, which might break some "FREEZE" and "TO_CBOR" methods |
250 |
that rely on this, such as bignum encoding, so this option is mainly |
251 |
useful for very simple data. |
252 |
|
253 |
$cbor = $cbor->validate_utf8 ([$enable]) |
254 |
$enabled = $cbor->get_validate_utf8 |
255 |
If $enable is true (or missing), then "decode" will validate that |
256 |
elements (text strings) containing UTF-8 data in fact contain valid |
257 |
UTF-8 data (instead of blindly accepting it). This validation |
258 |
obviously takes extra time during decoding. |
259 |
|
260 |
The concept of "valid UTF-8" used is perl's concept, which is a |
261 |
superset of the official UTF-8. |
262 |
|
263 |
If $enable is false (the default), then "decode" will blindly accept |
264 |
UTF-8 data, marking them as valid UTF-8 in the resulting data |
265 |
structure regardless of whether that's true or not. |
266 |
|
267 |
Perl isn't too happy about corrupted UTF-8 in strings, but should |
268 |
generally not crash or do similarly evil things. Extensions might be |
269 |
not so forgiving, so it's recommended to turn on this setting if you |
270 |
receive untrusted CBOR. |
271 |
|
272 |
This option does not affect "encode" in any way - strings that are |
273 |
supposedly valid UTF-8 will simply be dumped into the resulting CBOR |
274 |
string without checking whether that is, in fact, true or not. |
275 |
|
276 |
$cbor = $cbor->filter ([$cb->($tag, $value)]) |
277 |
$cb_or_undef = $cbor->get_filter |
278 |
Sets or replaces the tagged value decoding filter (when $cb is |
279 |
specified) or clears the filter (if no argument or "undef" is |
280 |
provided). |
281 |
|
282 |
The filter callback is called only during decoding, when a |
283 |
non-enforced tagged value has been decoded (see "TAG HANDLING AND |
284 |
EXTENSIONS" for a list of enforced tags). For specific tags, it's |
285 |
often better to provide a default converter using the |
286 |
%CBOR::XS::FILTER hash (see below). |
287 |
|
288 |
The first argument is the numerical tag, the second is the (decoded) |
289 |
value that has been tagged. |
290 |
|
291 |
The filter function should return either exactly one value, which |
292 |
will replace the tagged value in the decoded data structure, or no |
293 |
values, which will result in default handling, which currently means |
294 |
the decoder creates a "CBOR::XS::Tagged" object to hold the tag and |
295 |
the value. |
296 |
|
297 |
When the filter is cleared (the default state), the default filter |
298 |
function, "CBOR::XS::default_filter", is used. This function simply |
299 |
looks up the tag in the %CBOR::XS::FILTER hash. If an entry exists |
300 |
it must be a code reference that is called with tag and value, and |
301 |
is responsible for decoding the value. If no entry exists, it |
302 |
returns no values. |
303 |
|
304 |
Example: decode all tags not handled internally into |
305 |
"CBOR::XS::Tagged" objects, with no other special handling (useful |
306 |
when working with potentially "unsafe" CBOR data). |
307 |
|
308 |
CBOR::XS->new->filter (sub { })->decode ($cbor_data); |
309 |
|
310 |
Example: provide a global filter for tag 1347375694, converting the |
311 |
value into some string form. |
312 |
|
313 |
$CBOR::XS::FILTER{1347375694} = sub { |
314 |
my ($tag, $value); |
315 |
|
316 |
"tag 1347375694 value $value" |
317 |
}; |
318 |
|
319 |
$cbor_data = $cbor->encode ($perl_scalar) |
320 |
Converts the given Perl data structure (a scalar value) to its CBOR |
321 |
representation. |
322 |
|
323 |
$perl_scalar = $cbor->decode ($cbor_data) |
324 |
The opposite of "encode": expects CBOR data and tries to parse it, |
325 |
returning the resulting simple scalar or reference. Croaks on error. |
326 |
|
327 |
($perl_scalar, $octets) = $cbor->decode_prefix ($cbor_data) |
328 |
This works like the "decode" method, but instead of raising an |
329 |
exception when there is trailing garbage after the CBOR string, it |
330 |
will silently stop parsing there and return the number of characters |
331 |
consumed so far. |
332 |
|
333 |
This is useful if your CBOR texts are not delimited by an outer |
334 |
protocol and you need to know where the first CBOR string ends amd |
335 |
the next one starts. |
336 |
|
337 |
CBOR::XS->new->decode_prefix ("......") |
338 |
=> ("...", 3) |
339 |
|
340 |
INCREMENTAL PARSING |
341 |
In some cases, there is the need for incremental parsing of JSON texts. |
342 |
While this module always has to keep both CBOR text and resulting Perl |
343 |
data structure in memory at one time, it does allow you to parse a CBOR |
344 |
stream incrementally, using a similar to using "decode_prefix" to see if |
345 |
a full CBOR object is available, but is much more efficient. |
346 |
|
347 |
It basically works by parsing as much of a CBOR string as possible - if |
348 |
the CBOR data is not complete yet, the pasrer will remember where it |
349 |
was, to be able to restart when more data has been accumulated. Once |
350 |
enough data is available to either decode a complete CBOR value or raise |
351 |
an error, a real decode will be attempted. |
352 |
|
353 |
A typical use case would be a network protocol that consists of sending |
354 |
and receiving CBOR-encoded messages. The solution that works with CBOR |
355 |
and about anything else is by prepending a length to every CBOR value, |
356 |
so the receiver knows how many octets to read. More compact (and |
357 |
slightly slower) would be to just send CBOR values back-to-back, as |
358 |
"CBOR::XS" knows where a CBOR value ends, and doesn't need an explicit |
359 |
length. |
360 |
|
361 |
The following methods help with this: |
362 |
|
363 |
@decoded = $cbor->incr_parse ($buffer) |
364 |
This method attempts to decode exactly one CBOR value from the |
365 |
beginning of the given $buffer. The value is removed from the |
366 |
$buffer on success. When $buffer doesn't contain a complete value |
367 |
yet, it returns nothing. Finally, when the $buffer doesn't start |
368 |
with something that could ever be a valid CBOR value, it raises an |
369 |
exception, just as "decode" would. In the latter case the decoder |
370 |
state is undefined and must be reset before being able to parse |
371 |
further. |
372 |
|
373 |
This method modifies the $buffer in place. When no CBOR value can be |
374 |
decoded, the decoder stores the current string offset. On the next |
375 |
call, continues decoding at the place where it stopped before. For |
376 |
this to make sense, the $buffer must begin with the same octets as |
377 |
on previous unsuccessful calls. |
378 |
|
379 |
You can call this method in scalar context, in which case it either |
380 |
returns a decoded value or "undef". This makes it impossible to |
381 |
distinguish between CBOR null values (which decode to "undef") and |
382 |
an unsuccessful decode, which is often acceptable. |
383 |
|
384 |
@decoded = $cbor->incr_parse_multiple ($buffer) |
385 |
Same as "incr_parse", but attempts to decode as many CBOR values as |
386 |
possible in one go, instead of at most one. Calls to "incr_parse" |
387 |
and "incr_parse_multiple" can be interleaved. |
388 |
|
389 |
$cbor->incr_reset |
390 |
Resets the incremental decoder. This throws away any saved state, so |
391 |
that subsequent calls to "incr_parse" or "incr_parse_multiple" start |
392 |
to parse a new CBOR value from the beginning of the $buffer again. |
393 |
|
394 |
This method can be caled at any time, but it *must* be called if you |
395 |
want to change your $buffer or there was a decoding error and you |
396 |
want to reuse the $cbor object for future incremental parsings. |
397 |
|
398 |
MAPPING |
399 |
This section describes how CBOR::XS maps Perl values to CBOR values and |
400 |
vice versa. These mappings are designed to "do the right thing" in most |
401 |
circumstances automatically, preserving round-tripping characteristics |
402 |
(what you put in comes out as something equivalent). |
403 |
|
404 |
For the more enlightened: note that in the following descriptions, |
405 |
lowercase *perl* refers to the Perl interpreter, while uppercase *Perl* |
406 |
refers to the abstract Perl language itself. |
407 |
|
408 |
CBOR -> PERL |
409 |
integers |
410 |
CBOR integers become (numeric) perl scalars. On perls without 64 bit |
411 |
support, 64 bit integers will be truncated or otherwise corrupted. |
412 |
|
413 |
byte strings |
414 |
Byte strings will become octet strings in Perl (the Byte values |
415 |
0..255 will simply become characters of the same value in Perl). |
416 |
|
417 |
UTF-8 strings |
418 |
UTF-8 strings in CBOR will be decoded, i.e. the UTF-8 octets will be |
419 |
decoded into proper Unicode code points. At the moment, the validity |
420 |
of the UTF-8 octets will not be validated - corrupt input will |
421 |
result in corrupted Perl strings. |
422 |
|
423 |
arrays, maps |
424 |
CBOR arrays and CBOR maps will be converted into references to a |
425 |
Perl array or hash, respectively. The keys of the map will be |
426 |
stringified during this process. |
427 |
|
428 |
null |
429 |
CBOR null becomes "undef" in Perl. |
430 |
|
431 |
true, false, undefined |
432 |
These CBOR values become "Types:Serialiser::true", |
433 |
"Types:Serialiser::false" and "Types::Serialiser::error", |
434 |
respectively. They are overloaded to act almost exactly like the |
435 |
numbers 1 and 0 (for true and false) or to throw an exception on |
436 |
access (for error). See the Types::Serialiser manpage for details. |
437 |
|
438 |
tagged values |
439 |
Tagged items consists of a numeric tag and another CBOR value. |
440 |
|
441 |
See "TAG HANDLING AND EXTENSIONS" and the description of "->filter" |
442 |
for details on which tags are handled how. |
443 |
|
444 |
anything else |
445 |
Anything else (e.g. unsupported simple values) will raise a decoding |
446 |
error. |
447 |
|
448 |
PERL -> CBOR |
449 |
The mapping from Perl to CBOR is slightly more difficult, as Perl is a |
450 |
typeless language. That means this module can only guess which CBOR type |
451 |
is meant by a perl value. |
452 |
|
453 |
hash references |
454 |
Perl hash references become CBOR maps. As there is no inherent |
455 |
ordering in hash keys (or CBOR maps), they will usually be encoded |
456 |
in a pseudo-random order. This order can be different each time a |
457 |
hash is encoded. |
458 |
|
459 |
Currently, tied hashes will use the indefinite-length format, while |
460 |
normal hashes will use the fixed-length format. |
461 |
|
462 |
array references |
463 |
Perl array references become fixed-length CBOR arrays. |
464 |
|
465 |
other references |
466 |
Other unblessed references will be represented using the indirection |
467 |
tag extension (tag value 22098, |
468 |
<http://cbor.schmorp.de/indirection>). CBOR decoders are guaranteed |
469 |
to be able to decode these values somehow, by either "doing the |
470 |
right thing", decoding into a generic tagged object, simply ignoring |
471 |
the tag, or something else. |
472 |
|
473 |
CBOR::XS::Tagged objects |
474 |
Objects of this type must be arrays consisting of a single "[tag, |
475 |
value]" pair. The (numerical) tag will be encoded as a CBOR tag, the |
476 |
value will be encoded as appropriate for the value. You must use |
477 |
"CBOR::XS::tag" to create such objects. |
478 |
|
479 |
Types::Serialiser::true, Types::Serialiser::false, |
480 |
Types::Serialiser::error |
481 |
These special values become CBOR true, CBOR false and CBOR undefined |
482 |
values, respectively. You can also use "\1", "\0" and "\undef" |
483 |
directly if you want. |
484 |
|
485 |
other blessed objects |
486 |
Other blessed objects are serialised via "TO_CBOR" or "FREEZE". See |
487 |
"TAG HANDLING AND EXTENSIONS" for specific classes handled by this |
488 |
module, and "OBJECT SERIALISATION" for generic object serialisation. |
489 |
|
490 |
simple scalars |
491 |
Simple Perl scalars (any scalar that is not a reference) are the |
492 |
most difficult objects to encode: CBOR::XS will encode undefined |
493 |
scalars as CBOR null values, scalars that have last been used in a |
494 |
string context before encoding as CBOR strings, and anything else as |
495 |
number value: |
496 |
|
497 |
# dump as number |
498 |
encode_cbor [2] # yields [2] |
499 |
encode_cbor [-3.0e17] # yields [-3e+17] |
500 |
my $value = 5; encode_cbor [$value] # yields [5] |
501 |
|
502 |
# used as string, so dump as string (either byte or text) |
503 |
print $value; |
504 |
encode_cbor [$value] # yields ["5"] |
505 |
|
506 |
# undef becomes null |
507 |
encode_cbor [undef] # yields [null] |
508 |
|
509 |
You can force the type to be a CBOR string by stringifying it: |
510 |
|
511 |
my $x = 3.1; # some variable containing a number |
512 |
"$x"; # stringified |
513 |
$x .= ""; # another, more awkward way to stringify |
514 |
print $x; # perl does it for you, too, quite often |
515 |
|
516 |
You can force whether a string is encoded as byte or text string by |
517 |
using "utf8::upgrade" and "utf8::downgrade" (if "text_strings" is |
518 |
disabled): |
519 |
|
520 |
utf8::upgrade $x; # encode $x as text string |
521 |
utf8::downgrade $x; # encode $x as byte string |
522 |
|
523 |
Perl doesn't define what operations up- and downgrade strings, so if |
524 |
the difference between byte and text is important, you should up- or |
525 |
downgrade your string as late as possible before encoding. You can |
526 |
also force the use of CBOR text strings by using "text_keys" or |
527 |
"text_strings". |
528 |
|
529 |
You can force the type to be a CBOR number by numifying it: |
530 |
|
531 |
my $x = "3"; # some variable containing a string |
532 |
$x += 0; # numify it, ensuring it will be dumped as a number |
533 |
$x *= 1; # same thing, the choice is yours. |
534 |
|
535 |
You can not currently force the type in other, less obscure, ways. |
536 |
Tell me if you need this capability (but don't forget to explain why |
537 |
it's needed :). |
538 |
|
539 |
Perl values that seem to be integers generally use the shortest |
540 |
possible representation. Floating-point values will use either the |
541 |
IEEE single format if possible without loss of precision, otherwise |
542 |
the IEEE double format will be used. Perls that use formats other |
543 |
than IEEE double to represent numerical values are supported, but |
544 |
might suffer loss of precision. |
545 |
|
546 |
OBJECT SERIALISATION |
547 |
This module implements both a CBOR-specific and the generic |
548 |
Types::Serialier object serialisation protocol. The following |
549 |
subsections explain both methods. |
550 |
|
551 |
ENCODING |
552 |
This module knows two way to serialise a Perl object: The CBOR-specific |
553 |
way, and the generic way. |
554 |
|
555 |
Whenever the encoder encounters a Perl object that it cannot serialise |
556 |
directly (most of them), it will first look up the "TO_CBOR" method on |
557 |
it. |
558 |
|
559 |
If it has a "TO_CBOR" method, it will call it with the object as only |
560 |
argument, and expects exactly one return value, which it will then |
561 |
substitute and encode it in the place of the object. |
562 |
|
563 |
Otherwise, it will look up the "FREEZE" method. If it exists, it will |
564 |
call it with the object as first argument, and the constant string |
565 |
"CBOR" as the second argument, to distinguish it from other serialisers. |
566 |
|
567 |
The "FREEZE" method can return any number of values (i.e. zero or more). |
568 |
These will be encoded as CBOR perl object, together with the classname. |
569 |
|
570 |
These methods *MUST NOT* change the data structure that is being |
571 |
serialised. Failure to comply to this can result in memory corruption - |
572 |
and worse. |
573 |
|
574 |
If an object supports neither "TO_CBOR" nor "FREEZE", encoding will fail |
575 |
with an error. |
576 |
|
577 |
DECODING |
578 |
Objects encoded via "TO_CBOR" cannot (normally) be automatically |
579 |
decoded, but objects encoded via "FREEZE" can be decoded using the |
580 |
following protocol: |
581 |
|
582 |
When an encoded CBOR perl object is encountered by the decoder, it will |
583 |
look up the "THAW" method, by using the stored classname, and will fail |
584 |
if the method cannot be found. |
585 |
|
586 |
After the lookup it will call the "THAW" method with the stored |
587 |
classname as first argument, the constant string "CBOR" as second |
588 |
argument, and all values returned by "FREEZE" as remaining arguments. |
589 |
|
590 |
EXAMPLES |
591 |
Here is an example "TO_CBOR" method: |
592 |
|
593 |
sub My::Object::TO_CBOR { |
594 |
my ($obj) = @_; |
595 |
|
596 |
["this is a serialised My::Object object", $obj->{id}] |
597 |
} |
598 |
|
599 |
When a "My::Object" is encoded to CBOR, it will instead encode a simple |
600 |
array with two members: a string, and the "object id". Decoding this |
601 |
CBOR string will yield a normal perl array reference in place of the |
602 |
object. |
603 |
|
604 |
A more useful and practical example would be a serialisation method for |
605 |
the URI module. CBOR has a custom tag value for URIs, namely 32: |
606 |
|
607 |
sub URI::TO_CBOR { |
608 |
my ($self) = @_; |
609 |
my $uri = "$self"; # stringify uri |
610 |
utf8::upgrade $uri; # make sure it will be encoded as UTF-8 string |
611 |
CBOR::XS::tag 32, "$_[0]" |
612 |
} |
613 |
|
614 |
This will encode URIs as a UTF-8 string with tag 32, which indicates an |
615 |
URI. |
616 |
|
617 |
Decoding such an URI will not (currently) give you an URI object, but |
618 |
instead a CBOR::XS::Tagged object with tag number 32 and the string - |
619 |
exactly what was returned by "TO_CBOR". |
620 |
|
621 |
To serialise an object so it can automatically be deserialised, you need |
622 |
to use "FREEZE" and "THAW". To take the URI module as example, this |
623 |
would be a possible implementation: |
624 |
|
625 |
sub URI::FREEZE { |
626 |
my ($self, $serialiser) = @_; |
627 |
"$self" # encode url string |
628 |
} |
629 |
|
630 |
sub URI::THAW { |
631 |
my ($class, $serialiser, $uri) = @_; |
632 |
$class->new ($uri) |
633 |
} |
634 |
|
635 |
Unlike "TO_CBOR", multiple values can be returned by "FREEZE". For |
636 |
example, a "FREEZE" method that returns "type", "id" and "variant" |
637 |
values would cause an invocation of "THAW" with 5 arguments: |
638 |
|
639 |
sub My::Object::FREEZE { |
640 |
my ($self, $serialiser) = @_; |
641 |
|
642 |
($self->{type}, $self->{id}, $self->{variant}) |
643 |
} |
644 |
|
645 |
sub My::Object::THAW { |
646 |
my ($class, $serialiser, $type, $id, $variant) = @_; |
647 |
|
648 |
$class-<new (type => $type, id => $id, variant => $variant) |
649 |
} |
650 |
|
651 |
MAGIC HEADER |
652 |
There is no way to distinguish CBOR from other formats programmatically. |
653 |
To make it easier to distinguish CBOR from other formats, the CBOR |
654 |
specification has a special "magic string" that can be prepended to any |
655 |
CBOR string without changing its meaning. |
656 |
|
657 |
This string is available as $CBOR::XS::MAGIC. This module does not |
658 |
prepend this string to the CBOR data it generates, but it will ignore it |
659 |
if present, so users can prepend this string as a "file type" indicator |
660 |
as required. |
661 |
|
662 |
THE CBOR::XS::Tagged CLASS |
663 |
CBOR has the concept of tagged values - any CBOR value can be tagged |
664 |
with a numeric 64 bit number, which are centrally administered. |
665 |
|
666 |
"CBOR::XS" handles a few tags internally when en- or decoding. You can |
667 |
also create tags yourself by encoding "CBOR::XS::Tagged" objects, and |
668 |
the decoder will create "CBOR::XS::Tagged" objects itself when it hits |
669 |
an unknown tag. |
670 |
|
671 |
These objects are simply blessed array references - the first member of |
672 |
the array being the numerical tag, the second being the value. |
673 |
|
674 |
You can interact with "CBOR::XS::Tagged" objects in the following ways: |
675 |
|
676 |
$tagged = CBOR::XS::tag $tag, $value |
677 |
This function(!) creates a new "CBOR::XS::Tagged" object using the |
678 |
given $tag (0..2**64-1) to tag the given $value (which can be any |
679 |
Perl value that can be encoded in CBOR, including serialisable Perl |
680 |
objects and "CBOR::XS::Tagged" objects). |
681 |
|
682 |
$tagged->[0] |
683 |
$tagged->[0] = $new_tag |
684 |
$tag = $tagged->tag |
685 |
$new_tag = $tagged->tag ($new_tag) |
686 |
Access/mutate the tag. |
687 |
|
688 |
$tagged->[1] |
689 |
$tagged->[1] = $new_value |
690 |
$value = $tagged->value |
691 |
$new_value = $tagged->value ($new_value) |
692 |
Access/mutate the tagged value. |
693 |
|
694 |
EXAMPLES |
695 |
Here are some examples of "CBOR::XS::Tagged" uses to tag objects. |
696 |
|
697 |
You can look up CBOR tag value and emanings in the IANA registry at |
698 |
<http://www.iana.org/assignments/cbor-tags/cbor-tags.xhtml>. |
699 |
|
700 |
Prepend a magic header ($CBOR::XS::MAGIC): |
701 |
|
702 |
my $cbor = encode_cbor CBOR::XS::tag 55799, $value; |
703 |
# same as: |
704 |
my $cbor = $CBOR::XS::MAGIC . encode_cbor $value; |
705 |
|
706 |
Serialise some URIs and a regex in an array: |
707 |
|
708 |
my $cbor = encode_cbor [ |
709 |
(CBOR::XS::tag 32, "http://www.nethype.de/"), |
710 |
(CBOR::XS::tag 32, "http://software.schmorp.de/"), |
711 |
(CBOR::XS::tag 35, "^[Pp][Ee][Rr][lL]\$"), |
712 |
]; |
713 |
|
714 |
Wrap CBOR data in CBOR: |
715 |
|
716 |
my $cbor_cbor = encode_cbor |
717 |
CBOR::XS::tag 24, |
718 |
encode_cbor [1, 2, 3]; |
719 |
|
720 |
TAG HANDLING AND EXTENSIONS |
721 |
This section describes how this module handles specific tagged values |
722 |
and extensions. If a tag is not mentioned here and no additional filters |
723 |
are provided for it, then the default handling applies (creating a |
724 |
CBOR::XS::Tagged object on decoding, and only encoding the tag when |
725 |
explicitly requested). |
726 |
|
727 |
Tags not handled specifically are currently converted into a |
728 |
CBOR::XS::Tagged object, which is simply a blessed array reference |
729 |
consisting of the numeric tag value followed by the (decoded) CBOR |
730 |
value. |
731 |
|
732 |
Future versions of this module reserve the right to special case |
733 |
additional tags (such as base64url). |
734 |
|
735 |
ENFORCED TAGS |
736 |
These tags are always handled when decoding, and their handling cannot |
737 |
be overridden by the user. |
738 |
|
739 |
26 (perl-object, <http://cbor.schmorp.de/perl-object>) |
740 |
These tags are automatically created (and decoded) for serialisable |
741 |
objects using the "FREEZE/THAW" methods (the Types::Serialier object |
742 |
serialisation protocol). See "OBJECT SERIALISATION" for details. |
743 |
|
744 |
28, 29 (shareable, sharedref, <http://cbor.schmorp.de/value-sharing>) |
745 |
These tags are automatically decoded when encountered (and they do |
746 |
not result in a cyclic data structure, see "allow_cycles"), |
747 |
resulting in shared values in the decoded object. They are only |
748 |
encoded, however, when "allow_sharing" is enabled. |
749 |
|
750 |
Not all shared values can be successfully decoded: values that |
751 |
reference themselves will *currently* decode as "undef" (this is not |
752 |
the same as a reference pointing to itself, which will be |
753 |
represented as a value that contains an indirect reference to itself |
754 |
- these will be decoded properly). |
755 |
|
756 |
Note that considerably more shared value data structures can be |
757 |
decoded than will be encoded - currently, only values pointed to by |
758 |
references will be shared, others will not. While non-reference |
759 |
shared values can be generated in Perl with some effort, they were |
760 |
considered too unimportant to be supported in the encoder. The |
761 |
decoder, however, will decode these values as shared values. |
762 |
|
763 |
256, 25 (stringref-namespace, stringref, |
764 |
<http://cbor.schmorp.de/stringref>) |
765 |
These tags are automatically decoded when encountered. They are only |
766 |
encoded, however, when "pack_strings" is enabled. |
767 |
|
768 |
22098 (indirection, <http://cbor.schmorp.de/indirection>) |
769 |
This tag is automatically generated when a reference are encountered |
770 |
(with the exception of hash and array references). It is converted |
771 |
to a reference when decoding. |
772 |
|
773 |
55799 (self-describe CBOR, RFC 7049) |
774 |
This value is not generated on encoding (unless explicitly requested |
775 |
by the user), and is simply ignored when decoding. |
776 |
|
777 |
NON-ENFORCED TAGS |
778 |
These tags have default filters provided when decoding. Their handling |
779 |
can be overridden by changing the %CBOR::XS::FILTER entry for the tag, |
780 |
or by providing a custom "filter" callback when decoding. |
781 |
|
782 |
When they result in decoding into a specific Perl class, the module |
783 |
usually provides a corresponding "TO_CBOR" method as well. |
784 |
|
785 |
When any of these need to load additional modules that are not part of |
786 |
the perl core distribution (e.g. URI), it is (currently) up to the user |
787 |
to provide these modules. The decoding usually fails with an exception |
788 |
if the required module cannot be loaded. |
789 |
|
790 |
0, 1 (date/time string, seconds since the epoch) |
791 |
These tags are decoded into Time::Piece objects. The corresponding |
792 |
"Time::Piece::TO_CBOR" method always encodes into tag 1 values |
793 |
currently. |
794 |
|
795 |
The Time::Piece API is generally surprisingly bad, and fractional |
796 |
seconds are only accidentally kept intact, so watch out. On the plus |
797 |
side, the module comes with perl since 5.10, which has to count for |
798 |
something. |
799 |
|
800 |
2, 3 (positive/negative bignum) |
801 |
These tags are decoded into Math::BigInt objects. The corresponding |
802 |
"Math::BigInt::TO_CBOR" method encodes "small" bigints into normal |
803 |
CBOR integers, and others into positive/negative CBOR bignums. |
804 |
|
805 |
4, 5, 264, 265 (decimal fraction/bigfloat) |
806 |
Both decimal fractions and bigfloats are decoded into Math::BigFloat |
807 |
objects. The corresponding "Math::BigFloat::TO_CBOR" method *always* |
808 |
encodes into a decimal fraction (either tag 4 or 264). |
809 |
|
810 |
NaN and infinities are not encoded properly, as they cannot be |
811 |
represented in CBOR. |
812 |
|
813 |
See "BIGNUM SECURITY CONSIDERATIONS" for more info. |
814 |
|
815 |
30 (rational numbers) |
816 |
These tags are decoded into Math::BigRat objects. The corresponding |
817 |
"Math::BigRat::TO_CBOR" method encodes rational numbers with |
818 |
denominator 1 via their numerator only, i.e., they become normal |
819 |
integers or "bignums". |
820 |
|
821 |
See "BIGNUM SECURITY CONSIDERATIONS" for more info. |
822 |
|
823 |
21, 22, 23 (expected later JSON conversion) |
824 |
CBOR::XS is not a CBOR-to-JSON converter, and will simply ignore |
825 |
these tags. |
826 |
|
827 |
32 (URI) |
828 |
These objects decode into URI objects. The corresponding |
829 |
"URI::TO_CBOR" method again results in a CBOR URI value. |
830 |
|
831 |
CBOR and JSON |
832 |
CBOR is supposed to implement a superset of the JSON data model, and is, |
833 |
with some coercion, able to represent all JSON texts (something that |
834 |
other "binary JSON" formats such as BSON generally do not support). |
835 |
|
836 |
CBOR implements some extra hints and support for JSON interoperability, |
837 |
and the spec offers further guidance for conversion between CBOR and |
838 |
JSON. None of this is currently implemented in CBOR, and the guidelines |
839 |
in the spec do not result in correct round-tripping of data. If JSON |
840 |
interoperability is improved in the future, then the goal will be to |
841 |
ensure that decoded JSON data will round-trip encoding and decoding to |
842 |
CBOR intact. |
843 |
|
844 |
SECURITY CONSIDERATIONS |
845 |
When you are using CBOR in a protocol, talking to untrusted potentially |
846 |
hostile creatures requires relatively few measures. |
847 |
|
848 |
First of all, your CBOR decoder should be secure, that is, should not |
849 |
have any buffer overflows. Obviously, this module should ensure that and |
850 |
I am trying hard on making that true, but you never know. |
851 |
|
852 |
Second, you need to avoid resource-starving attacks. That means you |
853 |
should limit the size of CBOR data you accept, or make sure then when |
854 |
your resources run out, that's just fine (e.g. by using a separate |
855 |
process that can crash safely). The size of a CBOR string in octets is |
856 |
usually a good indication of the size of the resources required to |
857 |
decode it into a Perl structure. While CBOR::XS can check the size of |
858 |
the CBOR text, it might be too late when you already have it in memory, |
859 |
so you might want to check the size before you accept the string. |
860 |
|
861 |
Third, CBOR::XS recurses using the C stack when decoding objects and |
862 |
arrays. The C stack is a limited resource: for instance, on my amd64 |
863 |
machine with 8MB of stack size I can decode around 180k nested arrays |
864 |
but only 14k nested CBOR objects (due to perl itself recursing deeply on |
865 |
croak to free the temporary). If that is exceeded, the program crashes. |
866 |
To be conservative, the default nesting limit is set to 512. If your |
867 |
process has a smaller stack, you should adjust this setting accordingly |
868 |
with the "max_depth" method. |
869 |
|
870 |
Something else could bomb you, too, that I forgot to think of. In that |
871 |
case, you get to keep the pieces. I am always open for hints, though... |
872 |
|
873 |
Also keep in mind that CBOR::XS might leak contents of your Perl data |
874 |
structures in its error messages, so when you serialise sensitive |
875 |
information you might want to make sure that exceptions thrown by |
876 |
CBOR::XS will not end up in front of untrusted eyes. |
877 |
|
878 |
BIGNUM SECURITY CONSIDERATIONS |
879 |
CBOR::XS provides a "TO_CBOR" method for both Math::BigInt and |
880 |
Math::BigFloat that tries to encode the number in the simplest possible |
881 |
way, that is, either a CBOR integer, a CBOR bigint/decimal fraction (tag |
882 |
4) or an arbitrary-exponent decimal fraction (tag 264). Rational numbers |
883 |
(Math::BigRat, tag 30) can also contain bignums as members. |
884 |
|
885 |
CBOR::XS will also understand base-2 bigfloat or arbitrary-exponent |
886 |
bigfloats (tags 5 and 265), but it will never generate these on its own. |
887 |
|
888 |
Using the built-in Math::BigInt::Calc support, encoding and decoding |
889 |
decimal fractions is generally fast. Decoding bigints can be slow for |
890 |
very big numbers (tens of thousands of digits, something that could |
891 |
potentially be caught by limiting the size of CBOR texts), and decoding |
892 |
bigfloats or arbitrary-exponent bigfloats can be *extremely* slow |
893 |
(minutes, decades) for large exponents (roughly 40 bit and longer). |
894 |
|
895 |
Additionally, Math::BigInt can take advantage of other bignum libraries, |
896 |
such as Math::GMP, which cannot handle big floats with large exponents, |
897 |
and might simply abort or crash your program, due to their code quality. |
898 |
|
899 |
This can be a concern if you want to parse untrusted CBOR. If it is, you |
900 |
might want to disable decoding of tag 2 (bigint) and 3 (negative bigint) |
901 |
types. You should also disable types 5 and 265, as these can be slow |
902 |
even without bigints. |
903 |
|
904 |
Disabling bigints will also partially or fully disable types that rely |
905 |
on them, e.g. rational numbers that use bignums. |
906 |
|
907 |
CBOR IMPLEMENTATION NOTES |
908 |
This section contains some random implementation notes. They do not |
909 |
describe guaranteed behaviour, but merely behaviour as-is implemented |
910 |
right now. |
911 |
|
912 |
64 bit integers are only properly decoded when Perl was built with 64 |
913 |
bit support. |
914 |
|
915 |
Strings and arrays are encoded with a definite length. Hashes as well, |
916 |
unless they are tied (or otherwise magical). |
917 |
|
918 |
Only the double data type is supported for NV data types - when Perl |
919 |
uses long double to represent floating point values, they might not be |
920 |
encoded properly. Half precision types are accepted, but not encoded. |
921 |
|
922 |
Strict mode and canonical mode are not implemented. |
923 |
|
924 |
LIMITATIONS ON PERLS WITHOUT 64-BIT INTEGER SUPPORT |
925 |
On perls that were built without 64 bit integer support (these are rare |
926 |
nowadays, even on 32 bit architectures, as all major Perl distributions |
927 |
are built with 64 bit integer support), support for any kind of 64 bit |
928 |
integer in CBOR is very limited - most likely, these 64 bit values will |
929 |
be truncated, corrupted, or otherwise not decoded correctly. This also |
930 |
includes string, array and map sizes that are stored as 64 bit integers. |
931 |
|
932 |
THREADS |
933 |
This module is *not* guaranteed to be thread safe and there are no plans |
934 |
to change this until Perl gets thread support (as opposed to the |
935 |
horribly slow so-called "threads" which are simply slow and bloated |
936 |
process simulations - use fork, it's *much* faster, cheaper, better). |
937 |
|
938 |
(It might actually work, but you have been warned). |
939 |
|
940 |
BUGS |
941 |
While the goal of this module is to be correct, that unfortunately does |
942 |
not mean it's bug-free, only that I think its design is bug-free. If you |
943 |
keep reporting bugs they will be fixed swiftly, though. |
944 |
|
945 |
Please refrain from using rt.cpan.org or any other bug reporting |
946 |
service. I put the contact address into my modules for a reason. |
947 |
|
948 |
SEE ALSO |
949 |
The JSON and JSON::XS modules that do similar, but human-readable, |
950 |
serialisation. |
951 |
|
952 |
The Types::Serialiser module provides the data model for true, false and |
953 |
error values. |
954 |
|
955 |
AUTHOR |
956 |
Marc Lehmann <schmorp@schmorp.de> |
957 |
http://home.schmorp.de/ |
958 |
|