ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/JSON-XS/XS.pm
(Generate patch)

Comparing JSON-XS/XS.pm (file contents):
Revision 1.20 by root, Sun Mar 25 00:47:42 2007 UTC vs.
Revision 1.116 by root, Tue Feb 17 23:41:20 2009 UTC

1=head1 NAME 1=head1 NAME
2 2
3JSON::XS - JSON serialising/deserialising, done correctly and fast 3JSON::XS - JSON serialising/deserialising, done correctly and fast
4 4
5=encoding utf-8
6
7JSON::XS - 正しくて高速な JSON シリアライザ/デシリアライザ
8 (http://fleur.hio.jp/perldoc/mix/lib/JSON/XS.html)
9
5=head1 SYNOPSIS 10=head1 SYNOPSIS
6 11
7 use JSON::XS; 12 use JSON::XS;
8 13
9 # exported functions, croak on error 14 # exported functions, they croak on error
15 # and expect/generate UTF-8
10 16
11 $utf8_encoded_json_text = to_json $perl_hash_or_arrayref; 17 $utf8_encoded_json_text = encode_json $perl_hash_or_arrayref;
12 $perl_hash_or_arrayref = from_json $utf8_encoded_json_text; 18 $perl_hash_or_arrayref = decode_json $utf8_encoded_json_text;
13 19
14 # oo-interface 20 # OO-interface
15 21
16 $coder = JSON::XS->new->ascii->pretty->allow_nonref; 22 $coder = JSON::XS->new->ascii->pretty->allow_nonref;
17 $pretty_printed_unencoded = $coder->encode ($perl_scalar); 23 $pretty_printed_unencoded = $coder->encode ($perl_scalar);
18 $perl_scalar = $coder->decode ($unicode_json_text); 24 $perl_scalar = $coder->decode ($unicode_json_text);
19 25
26 # Note that JSON version 2.0 and above will automatically use JSON::XS
27 # if available, at virtually no speed overhead either, so you should
28 # be able to just:
29
30 use JSON;
31
32 # and do the same things, except that you have a pure-perl fallback now.
33
20=head1 DESCRIPTION 34=head1 DESCRIPTION
21 35
22This module converts Perl data structures to JSON and vice versa. Its 36This module converts Perl data structures to JSON and vice versa. Its
23primary goal is to be I<correct> and its secondary goal is to be 37primary goal is to be I<correct> and its secondary goal is to be
24I<fast>. To reach the latter goal it was written in C. 38I<fast>. To reach the latter goal it was written in C.
39
40Beginning with version 2.0 of the JSON module, when both JSON and
41JSON::XS are installed, then JSON will fall back on JSON::XS (this can be
42overridden) with no overhead due to emulation (by inheriting constructor
43and methods). If JSON::XS is not available, it will fall back to the
44compatible JSON::PP module as backend, so using JSON instead of JSON::XS
45gives you a portable JSON API that can be fast when you need and doesn't
46require a C compiler when that is a problem.
25 47
26As this is the n-th-something JSON module on CPAN, what was the reason 48As this is the n-th-something JSON module on CPAN, what was the reason
27to write yet another JSON module? While it seems there are many JSON 49to write yet another JSON module? While it seems there are many JSON
28modules, none of them correctly handle all corner cases, and in most cases 50modules, none of them correctly handle all corner cases, and in most cases
29their maintainers are unresponsive, gone missing, or not listening to bug 51their maintainers are unresponsive, gone missing, or not listening to bug
30reports for other reasons. 52reports for other reasons.
31 53
32See COMPARISON, below, for a comparison to some other JSON modules.
33
34See MAPPING, below, on how JSON::XS maps perl values to JSON values and 54See MAPPING, below, on how JSON::XS maps perl values to JSON values and
35vice versa. 55vice versa.
36 56
37=head2 FEATURES 57=head2 FEATURES
38 58
39=over 4 59=over 4
40 60
41=item * correct handling of unicode issues 61=item * correct Unicode handling
42 62
43This module knows how to handle Unicode, and even documents how and when 63This module knows how to handle Unicode, documents how and when it does
44it does so. 64so, and even documents what "correct" means.
45 65
46=item * round-trip integrity 66=item * round-trip integrity
47 67
48When you serialise a perl data structure using only datatypes supported 68When you serialise a perl data structure using only data types supported
49by JSON, the deserialised data structure is identical on the Perl level. 69by JSON, the deserialised data structure is identical on the Perl level.
50(e.g. the string "2.0" doesn't suddenly become "2"). 70(e.g. the string "2.0" doesn't suddenly become "2" just because it looks
71like a number). There minor I<are> exceptions to this, read the MAPPING
72section below to learn about those.
51 73
52=item * strict checking of JSON correctness 74=item * strict checking of JSON correctness
53 75
54There is no guessing, no generating of illegal JSON texts by default, 76There is no guessing, no generating of illegal JSON texts by default,
55and only JSON is accepted as input by default (the latter is a security 77and only JSON is accepted as input by default (the latter is a security
56feature). 78feature).
57 79
58=item * fast 80=item * fast
59 81
60Compared to other JSON modules, this module compares favourably in terms 82Compared to other JSON modules and other serialisers such as Storable,
61of speed, too. 83this module usually compares favourably in terms of speed, too.
62 84
63=item * simple to use 85=item * simple to use
64 86
65This module has both a simple functional interface as well as an OO 87This module has both a simple functional interface as well as an object
66interface. 88oriented interface interface.
67 89
68=item * reasonably versatile output formats 90=item * reasonably versatile output formats
69 91
70You can choose between the most compact guarenteed single-line format 92You can choose between the most compact guaranteed-single-line format
71possible (nice for simple line-based protocols), a pure-ascii format (for 93possible (nice for simple line-based protocols), a pure-ASCII format
72when your transport is not 8-bit clean), or a pretty-printed format (for 94(for when your transport is not 8-bit clean, still supports the whole
73when you want to read that stuff). Or you can combine those features in 95Unicode range), or a pretty-printed format (for when you want to read that
74whatever way you like. 96stuff). Or you can combine those features in whatever way you like.
75 97
76=back 98=back
77 99
78=cut 100=cut
79 101
80package JSON::XS; 102package JSON::XS;
81 103
104no warnings;
82use strict; 105use strict;
83 106
84BEGIN {
85 our $VERSION = '0.7'; 107our $VERSION = '2.232';
86 our @ISA = qw(Exporter); 108our @ISA = qw(Exporter);
87 109
88 our @EXPORT = qw(to_json from_json); 110our @EXPORT = qw(encode_json decode_json to_json from_json);
89 require Exporter;
90 111
112sub to_json($) {
91 require XSLoader; 113 require Carp;
92 XSLoader::load JSON::XS::, $VERSION; 114 Carp::croak ("JSON::XS::to_json has been renamed to encode_json, either downgrade to pre-2.0 versions of JSON::XS or rename the call");
93} 115}
94 116
117sub from_json($) {
118 require Carp;
119 Carp::croak ("JSON::XS::from_json has been renamed to decode_json, either downgrade to pre-2.0 versions of JSON::XS or rename the call");
120}
121
122use Exporter;
123use XSLoader;
124
95=head1 FUNCTIONAL INTERFACE 125=head1 FUNCTIONAL INTERFACE
96 126
97The following convinience methods are provided by this module. They are 127The following convenience methods are provided by this module. They are
98exported by default: 128exported by default:
99 129
100=over 4 130=over 4
101 131
102=item $json_text = to_json $perl_scalar 132=item $json_text = encode_json $perl_scalar
103 133
104Converts the given Perl data structure (a simple scalar or a reference to 134Converts the given Perl data structure to a UTF-8 encoded, binary string
105a hash or array) to a UTF-8 encoded, binary string (that is, the string contains 135(that is, the string contains octets only). Croaks on error.
106octets only). Croaks on error.
107 136
108This function call is functionally identical to: 137This function call is functionally identical to:
109 138
110 $json_text = JSON::XS->new->utf8->encode ($perl_scalar) 139 $json_text = JSON::XS->new->utf8->encode ($perl_scalar)
111 140
112except being faster. 141Except being faster.
113 142
114=item $perl_scalar = from_json $json_text 143=item $perl_scalar = decode_json $json_text
115 144
116The opposite of C<to_json>: expects an UTF-8 (binary) string and tries to 145The opposite of C<encode_json>: expects an UTF-8 (binary) string and tries
117parse that as an UTF-8 encoded JSON text, returning the resulting simple 146to parse that as an UTF-8 encoded JSON text, returning the resulting
118scalar or reference. Croaks on error. 147reference. Croaks on error.
119 148
120This function call is functionally identical to: 149This function call is functionally identical to:
121 150
122 $perl_scalar = JSON::XS->new->utf8->decode ($json_text) 151 $perl_scalar = JSON::XS->new->utf8->decode ($json_text)
123 152
124except being faster. 153Except being faster.
154
155=item $is_boolean = JSON::XS::is_bool $scalar
156
157Returns true if the passed scalar represents either JSON::XS::true or
158JSON::XS::false, two constants that act like C<1> and C<0>, respectively
159and are used to represent JSON C<true> and C<false> values in Perl.
160
161See MAPPING, below, for more information on how JSON values are mapped to
162Perl.
125 163
126=back 164=back
165
166
167=head1 A FEW NOTES ON UNICODE AND PERL
168
169Since this often leads to confusion, here are a few very clear words on
170how Unicode works in Perl, modulo bugs.
171
172=over 4
173
174=item 1. Perl strings can store characters with ordinal values > 255.
175
176This enables you to store Unicode characters as single characters in a
177Perl string - very natural.
178
179=item 2. Perl does I<not> associate an encoding with your strings.
180
181... until you force it to, e.g. when matching it against a regex, or
182printing the scalar to a file, in which case Perl either interprets your
183string as locale-encoded text, octets/binary, or as Unicode, depending
184on various settings. In no case is an encoding stored together with your
185data, it is I<use> that decides encoding, not any magical meta data.
186
187=item 3. The internal utf-8 flag has no meaning with regards to the
188encoding of your string.
189
190Just ignore that flag unless you debug a Perl bug, a module written in
191XS or want to dive into the internals of perl. Otherwise it will only
192confuse you, as, despite the name, it says nothing about how your string
193is encoded. You can have Unicode strings with that flag set, with that
194flag clear, and you can have binary data with that flag set and that flag
195clear. Other possibilities exist, too.
196
197If you didn't know about that flag, just the better, pretend it doesn't
198exist.
199
200=item 4. A "Unicode String" is simply a string where each character can be
201validly interpreted as a Unicode code point.
202
203If you have UTF-8 encoded data, it is no longer a Unicode string, but a
204Unicode string encoded in UTF-8, giving you a binary string.
205
206=item 5. A string containing "high" (> 255) character values is I<not> a UTF-8 string.
207
208It's a fact. Learn to live with it.
209
210=back
211
212I hope this helps :)
213
127 214
128=head1 OBJECT-ORIENTED INTERFACE 215=head1 OBJECT-ORIENTED INTERFACE
129 216
130The object oriented interface lets you configure your own encoding or 217The object oriented interface lets you configure your own encoding or
131decoding style, within the limits of supported formats. 218decoding style, within the limits of supported formats.
143 my $json = JSON::XS->new->utf8->space_after->encode ({a => [1,2]}) 230 my $json = JSON::XS->new->utf8->space_after->encode ({a => [1,2]})
144 => {"a": [1, 2]} 231 => {"a": [1, 2]}
145 232
146=item $json = $json->ascii ([$enable]) 233=item $json = $json->ascii ([$enable])
147 234
235=item $enabled = $json->get_ascii
236
148If C<$enable> is true (or missing), then the C<encode> method will not 237If C<$enable> is true (or missing), then the C<encode> method will not
149generate characters outside the code range C<0..127> (which is ASCII). Any 238generate characters outside the code range C<0..127> (which is ASCII). Any
150unicode characters outside that range will be escaped using either a 239Unicode characters outside that range will be escaped using either a
151single \uXXXX (BMP characters) or a double \uHHHH\uLLLLL escape sequence, 240single \uXXXX (BMP characters) or a double \uHHHH\uLLLLL escape sequence,
152as per RFC4627. 241as per RFC4627. The resulting encoded JSON text can be treated as a native
242Unicode string, an ascii-encoded, latin1-encoded or UTF-8 encoded string,
243or any other superset of ASCII.
153 244
154If C<$enable> is false, then the C<encode> method will not escape Unicode 245If C<$enable> is false, then the C<encode> method will not escape Unicode
155characters unless required by the JSON syntax. This results in a faster 246characters unless required by the JSON syntax or other flags. This results
156and more compact format. 247in a faster and more compact format.
248
249See also the section I<ENCODING/CODESET FLAG NOTES> later in this
250document.
251
252The main use for this flag is to produce JSON texts that can be
253transmitted over a 7-bit channel, as the encoded JSON texts will not
254contain any 8 bit characters.
157 255
158 JSON::XS->new->ascii (1)->encode ([chr 0x10401]) 256 JSON::XS->new->ascii (1)->encode ([chr 0x10401])
159 => ["\ud801\udc01"] 257 => ["\ud801\udc01"]
160 258
259=item $json = $json->latin1 ([$enable])
260
261=item $enabled = $json->get_latin1
262
263If C<$enable> is true (or missing), then the C<encode> method will encode
264the resulting JSON text as latin1 (or iso-8859-1), escaping any characters
265outside the code range C<0..255>. The resulting string can be treated as a
266latin1-encoded JSON text or a native Unicode string. The C<decode> method
267will not be affected in any way by this flag, as C<decode> by default
268expects Unicode, which is a strict superset of latin1.
269
270If C<$enable> is false, then the C<encode> method will not escape Unicode
271characters unless required by the JSON syntax or other flags.
272
273See also the section I<ENCODING/CODESET FLAG NOTES> later in this
274document.
275
276The main use for this flag is efficiently encoding binary data as JSON
277text, as most octets will not be escaped, resulting in a smaller encoded
278size. The disadvantage is that the resulting JSON text is encoded
279in latin1 (and must correctly be treated as such when storing and
280transferring), a rare encoding for JSON. It is therefore most useful when
281you want to store data structures known to contain binary data efficiently
282in files or databases, not when talking to other JSON encoders/decoders.
283
284 JSON::XS->new->latin1->encode (["\x{89}\x{abc}"]
285 => ["\x{89}\\u0abc"] # (perl syntax, U+abc escaped, U+89 not)
286
161=item $json = $json->utf8 ([$enable]) 287=item $json = $json->utf8 ([$enable])
288
289=item $enabled = $json->get_utf8
162 290
163If C<$enable> is true (or missing), then the C<encode> method will encode 291If C<$enable> is true (or missing), then the C<encode> method will encode
164the JSON result into UTF-8, as required by many protocols, while the 292the JSON result into UTF-8, as required by many protocols, while the
165C<decode> method expects to be handled an UTF-8-encoded string. Please 293C<decode> method expects to be handled an UTF-8-encoded string. Please
166note that UTF-8-encoded strings do not contain any characters outside the 294note that UTF-8-encoded strings do not contain any characters outside the
167range C<0..255>, they are thus useful for bytewise/binary I/O. In future 295range C<0..255>, they are thus useful for bytewise/binary I/O. In future
168versions, enabling this option might enable autodetection of the UTF-16 296versions, enabling this option might enable autodetection of the UTF-16
169and UTF-32 encoding families, as described in RFC4627. 297and UTF-32 encoding families, as described in RFC4627.
170 298
171If C<$enable> is false, then the C<encode> method will return the JSON 299If C<$enable> is false, then the C<encode> method will return the JSON
172string as a (non-encoded) unicode string, while C<decode> expects thus a 300string as a (non-encoded) Unicode string, while C<decode> expects thus a
173unicode string. Any decoding or encoding (e.g. to UTF-8 or UTF-16) needs 301Unicode string. Any decoding or encoding (e.g. to UTF-8 or UTF-16) needs
174to be done yourself, e.g. using the Encode module. 302to be done yourself, e.g. using the Encode module.
303
304See also the section I<ENCODING/CODESET FLAG NOTES> later in this
305document.
175 306
176Example, output UTF-16BE-encoded JSON: 307Example, output UTF-16BE-encoded JSON:
177 308
178 use Encode; 309 use Encode;
179 $jsontext = encode "UTF-16BE", JSON::XS->new->encode ($object); 310 $jsontext = encode "UTF-16BE", JSON::XS->new->encode ($object);
200 ] 331 ]
201 } 332 }
202 333
203=item $json = $json->indent ([$enable]) 334=item $json = $json->indent ([$enable])
204 335
336=item $enabled = $json->get_indent
337
205If C<$enable> is true (or missing), then the C<encode> method will use a multiline 338If C<$enable> is true (or missing), then the C<encode> method will use a multiline
206format as output, putting every array member or object/hash key-value pair 339format as output, putting every array member or object/hash key-value pair
207into its own line, identing them properly. 340into its own line, indenting them properly.
208 341
209If C<$enable> is false, no newlines or indenting will be produced, and the 342If C<$enable> is false, no newlines or indenting will be produced, and the
210resulting JSON text is guarenteed not to contain any C<newlines>. 343resulting JSON text is guaranteed not to contain any C<newlines>.
211 344
212This setting has no effect when decoding JSON texts. 345This setting has no effect when decoding JSON texts.
213 346
214=item $json = $json->space_before ([$enable]) 347=item $json = $json->space_before ([$enable])
348
349=item $enabled = $json->get_space_before
215 350
216If C<$enable> is true (or missing), then the C<encode> method will add an extra 351If C<$enable> is true (or missing), then the C<encode> method will add an extra
217optional space before the C<:> separating keys from values in JSON objects. 352optional space before the C<:> separating keys from values in JSON objects.
218 353
219If C<$enable> is false, then the C<encode> method will not add any extra 354If C<$enable> is false, then the C<encode> method will not add any extra
225Example, space_before enabled, space_after and indent disabled: 360Example, space_before enabled, space_after and indent disabled:
226 361
227 {"key" :"value"} 362 {"key" :"value"}
228 363
229=item $json = $json->space_after ([$enable]) 364=item $json = $json->space_after ([$enable])
365
366=item $enabled = $json->get_space_after
230 367
231If C<$enable> is true (or missing), then the C<encode> method will add an extra 368If C<$enable> is true (or missing), then the C<encode> method will add an extra
232optional space after the C<:> separating keys from values in JSON objects 369optional space after the C<:> separating keys from values in JSON objects
233and extra whitespace after the C<,> separating key-value pairs and array 370and extra whitespace after the C<,> separating key-value pairs and array
234members. 371members.
240 377
241Example, space_before and indent disabled, space_after enabled: 378Example, space_before and indent disabled, space_after enabled:
242 379
243 {"key": "value"} 380 {"key": "value"}
244 381
382=item $json = $json->relaxed ([$enable])
383
384=item $enabled = $json->get_relaxed
385
386If C<$enable> is true (or missing), then C<decode> will accept some
387extensions to normal JSON syntax (see below). C<encode> will not be
388affected in anyway. I<Be aware that this option makes you accept invalid
389JSON texts as if they were valid!>. I suggest only to use this option to
390parse application-specific files written by humans (configuration files,
391resource files etc.)
392
393If C<$enable> is false (the default), then C<decode> will only accept
394valid JSON texts.
395
396Currently accepted extensions are:
397
398=over 4
399
400=item * list items can have an end-comma
401
402JSON I<separates> array elements and key-value pairs with commas. This
403can be annoying if you write JSON texts manually and want to be able to
404quickly append elements, so this extension accepts comma at the end of
405such items not just between them:
406
407 [
408 1,
409 2, <- this comma not normally allowed
410 ]
411 {
412 "k1": "v1",
413 "k2": "v2", <- this comma not normally allowed
414 }
415
416=item * shell-style '#'-comments
417
418Whenever JSON allows whitespace, shell-style comments are additionally
419allowed. They are terminated by the first carriage-return or line-feed
420character, after which more white-space and comments are allowed.
421
422 [
423 1, # this comment not allowed in JSON
424 # neither this one...
425 ]
426
427=back
428
245=item $json = $json->canonical ([$enable]) 429=item $json = $json->canonical ([$enable])
430
431=item $enabled = $json->get_canonical
246 432
247If C<$enable> is true (or missing), then the C<encode> method will output JSON objects 433If C<$enable> is true (or missing), then the C<encode> method will output JSON objects
248by sorting their keys. This is adding a comparatively high overhead. 434by sorting their keys. This is adding a comparatively high overhead.
249 435
250If C<$enable> is false, then the C<encode> method will output key-value 436If C<$enable> is false, then the C<encode> method will output key-value
251pairs in the order Perl stores them (which will likely change between runs 437pairs in the order Perl stores them (which will likely change between runs
252of the same script). 438of the same script).
253 439
254This option is useful if you want the same data structure to be encoded as 440This option is useful if you want the same data structure to be encoded as
255the same JSON text (given the same overall settings). If it is disabled, 441the same JSON text (given the same overall settings). If it is disabled,
256the same hash migh be encoded differently even if contains the same data, 442the same hash might be encoded differently even if contains the same data,
257as key-value pairs have no inherent ordering in Perl. 443as key-value pairs have no inherent ordering in Perl.
258 444
259This setting has no effect when decoding JSON texts. 445This setting has no effect when decoding JSON texts.
260 446
261=item $json = $json->allow_nonref ([$enable]) 447=item $json = $json->allow_nonref ([$enable])
448
449=item $enabled = $json->get_allow_nonref
262 450
263If C<$enable> is true (or missing), then the C<encode> method can convert a 451If C<$enable> is true (or missing), then the C<encode> method can convert a
264non-reference into its corresponding string, number or null JSON value, 452non-reference into its corresponding string, number or null JSON value,
265which is an extension to RFC4627. Likewise, C<decode> will accept those JSON 453which is an extension to RFC4627. Likewise, C<decode> will accept those JSON
266values instead of croaking. 454values instead of croaking.
274resulting in an invalid JSON text: 462resulting in an invalid JSON text:
275 463
276 JSON::XS->new->allow_nonref->encode ("Hello, World!") 464 JSON::XS->new->allow_nonref->encode ("Hello, World!")
277 => "Hello, World!" 465 => "Hello, World!"
278 466
467=item $json = $json->allow_unknown ([$enable])
468
469=item $enabled = $json->get_allow_unknown
470
471If C<$enable> is true (or missing), then C<encode> will I<not> throw an
472exception when it encounters values it cannot represent in JSON (for
473example, filehandles) but instead will encode a JSON C<null> value. Note
474that blessed objects are not included here and are handled separately by
475c<allow_nonref>.
476
477If C<$enable> is false (the default), then C<encode> will throw an
478exception when it encounters anything it cannot encode as JSON.
479
480This option does not affect C<decode> in any way, and it is recommended to
481leave it off unless you know your communications partner.
482
483=item $json = $json->allow_blessed ([$enable])
484
485=item $enabled = $json->get_allow_blessed
486
487If C<$enable> is true (or missing), then the C<encode> method will not
488barf when it encounters a blessed reference. Instead, the value of the
489B<convert_blessed> option will decide whether C<null> (C<convert_blessed>
490disabled or no C<TO_JSON> method found) or a representation of the
491object (C<convert_blessed> enabled and C<TO_JSON> method found) is being
492encoded. Has no effect on C<decode>.
493
494If C<$enable> is false (the default), then C<encode> will throw an
495exception when it encounters a blessed object.
496
497=item $json = $json->convert_blessed ([$enable])
498
499=item $enabled = $json->get_convert_blessed
500
501If C<$enable> is true (or missing), then C<encode>, upon encountering a
502blessed object, will check for the availability of the C<TO_JSON> method
503on the object's class. If found, it will be called in scalar context
504and the resulting scalar will be encoded instead of the object. If no
505C<TO_JSON> method is found, the value of C<allow_blessed> will decide what
506to do.
507
508The C<TO_JSON> method may safely call die if it wants. If C<TO_JSON>
509returns other blessed objects, those will be handled in the same
510way. C<TO_JSON> must take care of not causing an endless recursion cycle
511(== crash) in this case. The name of C<TO_JSON> was chosen because other
512methods called by the Perl core (== not by the user of the object) are
513usually in upper case letters and to avoid collisions with any C<to_json>
514function or method.
515
516This setting does not yet influence C<decode> in any way, but in the
517future, global hooks might get installed that influence C<decode> and are
518enabled by this setting.
519
520If C<$enable> is false, then the C<allow_blessed> setting will decide what
521to do when a blessed object is found.
522
523=item $json = $json->filter_json_object ([$coderef->($hashref)])
524
525When C<$coderef> is specified, it will be called from C<decode> each
526time it decodes a JSON object. The only argument is a reference to the
527newly-created hash. If the code references returns a single scalar (which
528need not be a reference), this value (i.e. a copy of that scalar to avoid
529aliasing) is inserted into the deserialised data structure. If it returns
530an empty list (NOTE: I<not> C<undef>, which is a valid scalar), the
531original deserialised hash will be inserted. This setting can slow down
532decoding considerably.
533
534When C<$coderef> is omitted or undefined, any existing callback will
535be removed and C<decode> will not change the deserialised hash in any
536way.
537
538Example, convert all JSON objects into the integer 5:
539
540 my $js = JSON::XS->new->filter_json_object (sub { 5 });
541 # returns [5]
542 $js->decode ('[{}]')
543 # throw an exception because allow_nonref is not enabled
544 # so a lone 5 is not allowed.
545 $js->decode ('{"a":1, "b":2}');
546
547=item $json = $json->filter_json_single_key_object ($key [=> $coderef->($value)])
548
549Works remotely similar to C<filter_json_object>, but is only called for
550JSON objects having a single key named C<$key>.
551
552This C<$coderef> is called before the one specified via
553C<filter_json_object>, if any. It gets passed the single value in the JSON
554object. If it returns a single value, it will be inserted into the data
555structure. If it returns nothing (not even C<undef> but the empty list),
556the callback from C<filter_json_object> will be called next, as if no
557single-key callback were specified.
558
559If C<$coderef> is omitted or undefined, the corresponding callback will be
560disabled. There can only ever be one callback for a given key.
561
562As this callback gets called less often then the C<filter_json_object>
563one, decoding speed will not usually suffer as much. Therefore, single-key
564objects make excellent targets to serialise Perl objects into, especially
565as single-key JSON objects are as close to the type-tagged value concept
566as JSON gets (it's basically an ID/VALUE tuple). Of course, JSON does not
567support this in any way, so you need to make sure your data never looks
568like a serialised Perl hash.
569
570Typical names for the single object key are C<__class_whatever__>, or
571C<$__dollars_are_rarely_used__$> or C<}ugly_brace_placement>, or even
572things like C<__class_md5sum(classname)__>, to reduce the risk of clashing
573with real hashes.
574
575Example, decode JSON objects of the form C<< { "__widget__" => <id> } >>
576into the corresponding C<< $WIDGET{<id>} >> object:
577
578 # return whatever is in $WIDGET{5}:
579 JSON::XS
580 ->new
581 ->filter_json_single_key_object (__widget__ => sub {
582 $WIDGET{ $_[0] }
583 })
584 ->decode ('{"__widget__": 5')
585
586 # this can be used with a TO_JSON method in some "widget" class
587 # for serialisation to json:
588 sub WidgetBase::TO_JSON {
589 my ($self) = @_;
590
591 unless ($self->{id}) {
592 $self->{id} = ..get..some..id..;
593 $WIDGET{$self->{id}} = $self;
594 }
595
596 { __widget__ => $self->{id} }
597 }
598
279=item $json = $json->shrink ([$enable]) 599=item $json = $json->shrink ([$enable])
280 600
601=item $enabled = $json->get_shrink
602
281Perl usually over-allocates memory a bit when allocating space for 603Perl usually over-allocates memory a bit when allocating space for
282strings. This flag optionally resizes strings generated by either 604strings. This flag optionally resizes strings generated by either
283C<encode> or C<decode> to their minimum size possible. This can save 605C<encode> or C<decode> to their minimum size possible. This can save
284memory when your JSON texts are either very very long or you have many 606memory when your JSON texts are either very very long or you have many
285short strings. It will also try to downgrade any strings to octet-form 607short strings. It will also try to downgrade any strings to octet-form
286if possible: perl stores strings internally either in an encoding called 608if possible: perl stores strings internally either in an encoding called
287UTF-X or in octet-form. The latter cannot store everything but uses less 609UTF-X or in octet-form. The latter cannot store everything but uses less
288space in general. 610space in general (and some buggy Perl or C code might even rely on that
611internal representation being used).
289 612
613The actual definition of what shrink does might change in future versions,
614but it will always try to save space at the expense of time.
615
290If C<$enable> is true (or missing), the string returned by C<encode> will be shrunk-to-fit, 616If C<$enable> is true (or missing), the string returned by C<encode> will
291while all strings generated by C<decode> will also be shrunk-to-fit. 617be shrunk-to-fit, while all strings generated by C<decode> will also be
618shrunk-to-fit.
292 619
293If C<$enable> is false, then the normal perl allocation algorithms are used. 620If C<$enable> is false, then the normal perl allocation algorithms are used.
294If you work with your data, then this is likely to be faster. 621If you work with your data, then this is likely to be faster.
295 622
296In the future, this setting might control other things, such as converting 623In the future, this setting might control other things, such as converting
297strings that look like integers or floats into integers or floats 624strings that look like integers or floats into integers or floats
298internally (there is no difference on the Perl level), saving space. 625internally (there is no difference on the Perl level), saving space.
626
627=item $json = $json->max_depth ([$maximum_nesting_depth])
628
629=item $max_depth = $json->get_max_depth
630
631Sets the maximum nesting level (default C<512>) accepted while encoding
632or decoding. If a higher nesting level is detected in JSON text or a Perl
633data structure, then the encoder and decoder will stop and croak at that
634point.
635
636Nesting level is defined by number of hash- or arrayrefs that the encoder
637needs to traverse to reach a given point or the number of C<{> or C<[>
638characters without their matching closing parenthesis crossed to reach a
639given character in a string.
640
641Setting the maximum depth to one disallows any nesting, so that ensures
642that the object is only a single hash/object or array.
643
644If no argument is given, the highest possible setting will be used, which
645is rarely useful.
646
647Note that nesting is implemented by recursion in C. The default value has
648been chosen to be as large as typical operating systems allow without
649crashing.
650
651See SECURITY CONSIDERATIONS, below, for more info on why this is useful.
652
653=item $json = $json->max_size ([$maximum_string_size])
654
655=item $max_size = $json->get_max_size
656
657Set the maximum length a JSON text may have (in bytes) where decoding is
658being attempted. The default is C<0>, meaning no limit. When C<decode>
659is called on a string that is longer then this many bytes, it will not
660attempt to decode the string but throw an exception. This setting has no
661effect on C<encode> (yet).
662
663If no argument is given, the limit check will be deactivated (same as when
664C<0> is specified).
665
666See SECURITY CONSIDERATIONS, below, for more info on why this is useful.
299 667
300=item $json_text = $json->encode ($perl_scalar) 668=item $json_text = $json->encode ($perl_scalar)
301 669
302Converts the given Perl data structure (a simple scalar or a reference 670Converts the given Perl data structure (a simple scalar or a reference
303to a hash or array) to its JSON representation. Simple scalars will be 671to a hash or array) to its JSON representation. Simple scalars will be
313 681
314JSON numbers and strings become simple Perl scalars. JSON arrays become 682JSON numbers and strings become simple Perl scalars. JSON arrays become
315Perl arrayrefs and JSON objects become Perl hashrefs. C<true> becomes 683Perl arrayrefs and JSON objects become Perl hashrefs. C<true> becomes
316C<1>, C<false> becomes C<0> and C<null> becomes C<undef>. 684C<1>, C<false> becomes C<0> and C<null> becomes C<undef>.
317 685
686=item ($perl_scalar, $characters) = $json->decode_prefix ($json_text)
687
688This works like the C<decode> method, but instead of raising an exception
689when there is trailing garbage after the first JSON object, it will
690silently stop parsing there and return the number of characters consumed
691so far.
692
693This is useful if your JSON texts are not delimited by an outer protocol
694(which is not the brightest thing to do in the first place) and you need
695to know where the JSON text ends.
696
697 JSON::XS->new->decode_prefix ("[1] the tail")
698 => ([], 3)
699
318=back 700=back
701
702
703=head1 INCREMENTAL PARSING
704
705In some cases, there is the need for incremental parsing of JSON
706texts. While this module always has to keep both JSON text and resulting
707Perl data structure in memory at one time, it does allow you to parse a
708JSON stream incrementally. It does so by accumulating text until it has
709a full JSON object, which it then can decode. This process is similar to
710using C<decode_prefix> to see if a full JSON object is available, but
711is much more efficient (and can be implemented with a minimum of method
712calls).
713
714JSON::XS will only attempt to parse the JSON text once it is sure it
715has enough text to get a decisive result, using a very simple but
716truly incremental parser. This means that it sometimes won't stop as
717early as the full parser, for example, it doesn't detect parenthese
718mismatches. The only thing it guarantees is that it starts decoding as
719soon as a syntactically valid JSON text has been seen. This means you need
720to set resource limits (e.g. C<max_size>) to ensure the parser will stop
721parsing in the presence if syntax errors.
722
723The following methods implement this incremental parser.
724
725=over 4
726
727=item [void, scalar or list context] = $json->incr_parse ([$string])
728
729This is the central parsing function. It can both append new text and
730extract objects from the stream accumulated so far (both of these
731functions are optional).
732
733If C<$string> is given, then this string is appended to the already
734existing JSON fragment stored in the C<$json> object.
735
736After that, if the function is called in void context, it will simply
737return without doing anything further. This can be used to add more text
738in as many chunks as you want.
739
740If the method is called in scalar context, then it will try to extract
741exactly I<one> JSON object. If that is successful, it will return this
742object, otherwise it will return C<undef>. If there is a parse error,
743this method will croak just as C<decode> would do (one can then use
744C<incr_skip> to skip the errornous part). This is the most common way of
745using the method.
746
747And finally, in list context, it will try to extract as many objects
748from the stream as it can find and return them, or the empty list
749otherwise. For this to work, there must be no separators between the JSON
750objects or arrays, instead they must be concatenated back-to-back. If
751an error occurs, an exception will be raised as in the scalar context
752case. Note that in this case, any previously-parsed JSON texts will be
753lost.
754
755=item $lvalue_string = $json->incr_text
756
757This method returns the currently stored JSON fragment as an lvalue, that
758is, you can manipulate it. This I<only> works when a preceding call to
759C<incr_parse> in I<scalar context> successfully returned an object. Under
760all other circumstances you must not call this function (I mean it.
761although in simple tests it might actually work, it I<will> fail under
762real world conditions). As a special exception, you can also call this
763method before having parsed anything.
764
765This function is useful in two cases: a) finding the trailing text after a
766JSON object or b) parsing multiple JSON objects separated by non-JSON text
767(such as commas).
768
769=item $json->incr_skip
770
771This will reset the state of the incremental parser and will remove
772the parsed text from the input buffer so far. This is useful after
773C<incr_parse> died, in which case the input buffer and incremental parser
774state is left unchanged, to skip the text parsed so far and to reset the
775parse state.
776
777The difference to C<incr_reset> is that only text until the parse error
778occured is removed.
779
780=item $json->incr_reset
781
782This completely resets the incremental parser, that is, after this call,
783it will be as if the parser had never parsed anything.
784
785This is useful if you want to repeatedly parse JSON objects and want to
786ignore any trailing data, which means you have to reset the parser after
787each successful decode.
788
789=back
790
791=head2 LIMITATIONS
792
793All options that affect decoding are supported, except
794C<allow_nonref>. The reason for this is that it cannot be made to
795work sensibly: JSON objects and arrays are self-delimited, i.e. you can concatenate
796them back to back and still decode them perfectly. This does not hold true
797for JSON numbers, however.
798
799For example, is the string C<1> a single JSON number, or is it simply the
800start of C<12>? Or is C<12> a single JSON number, or the concatenation
801of C<1> and C<2>? In neither case you can tell, and this is why JSON::XS
802takes the conservative route and disallows this case.
803
804=head2 EXAMPLES
805
806Some examples will make all this clearer. First, a simple example that
807works similarly to C<decode_prefix>: We want to decode the JSON object at
808the start of a string and identify the portion after the JSON object:
809
810 my $text = "[1,2,3] hello";
811
812 my $json = new JSON::XS;
813
814 my $obj = $json->incr_parse ($text)
815 or die "expected JSON object or array at beginning of string";
816
817 my $tail = $json->incr_text;
818 # $tail now contains " hello"
819
820Easy, isn't it?
821
822Now for a more complicated example: Imagine a hypothetical protocol where
823you read some requests from a TCP stream, and each request is a JSON
824array, without any separation between them (in fact, it is often useful to
825use newlines as "separators", as these get interpreted as whitespace at
826the start of the JSON text, which makes it possible to test said protocol
827with C<telnet>...).
828
829Here is how you'd do it (it is trivial to write this in an event-based
830manner):
831
832 my $json = new JSON::XS;
833
834 # read some data from the socket
835 while (sysread $socket, my $buf, 4096) {
836
837 # split and decode as many requests as possible
838 for my $request ($json->incr_parse ($buf)) {
839 # act on the $request
840 }
841 }
842
843Another complicated example: Assume you have a string with JSON objects
844or arrays, all separated by (optional) comma characters (e.g. C<[1],[2],
845[3]>). To parse them, we have to skip the commas between the JSON texts,
846and here is where the lvalue-ness of C<incr_text> comes in useful:
847
848 my $text = "[1],[2], [3]";
849 my $json = new JSON::XS;
850
851 # void context, so no parsing done
852 $json->incr_parse ($text);
853
854 # now extract as many objects as possible. note the
855 # use of scalar context so incr_text can be called.
856 while (my $obj = $json->incr_parse) {
857 # do something with $obj
858
859 # now skip the optional comma
860 $json->incr_text =~ s/^ \s* , //x;
861 }
862
863Now lets go for a very complex example: Assume that you have a gigantic
864JSON array-of-objects, many gigabytes in size, and you want to parse it,
865but you cannot load it into memory fully (this has actually happened in
866the real world :).
867
868Well, you lost, you have to implement your own JSON parser. But JSON::XS
869can still help you: You implement a (very simple) array parser and let
870JSON decode the array elements, which are all full JSON objects on their
871own (this wouldn't work if the array elements could be JSON numbers, for
872example):
873
874 my $json = new JSON::XS;
875
876 # open the monster
877 open my $fh, "<bigfile.json"
878 or die "bigfile: $!";
879
880 # first parse the initial "["
881 for (;;) {
882 sysread $fh, my $buf, 65536
883 or die "read error: $!";
884 $json->incr_parse ($buf); # void context, so no parsing
885
886 # Exit the loop once we found and removed(!) the initial "[".
887 # In essence, we are (ab-)using the $json object as a simple scalar
888 # we append data to.
889 last if $json->incr_text =~ s/^ \s* \[ //x;
890 }
891
892 # now we have the skipped the initial "[", so continue
893 # parsing all the elements.
894 for (;;) {
895 # in this loop we read data until we got a single JSON object
896 for (;;) {
897 if (my $obj = $json->incr_parse) {
898 # do something with $obj
899 last;
900 }
901
902 # add more data
903 sysread $fh, my $buf, 65536
904 or die "read error: $!";
905 $json->incr_parse ($buf); # void context, so no parsing
906 }
907
908 # in this loop we read data until we either found and parsed the
909 # separating "," between elements, or the final "]"
910 for (;;) {
911 # first skip whitespace
912 $json->incr_text =~ s/^\s*//;
913
914 # if we find "]", we are done
915 if ($json->incr_text =~ s/^\]//) {
916 print "finished.\n";
917 exit;
918 }
919
920 # if we find ",", we can continue with the next element
921 if ($json->incr_text =~ s/^,//) {
922 last;
923 }
924
925 # if we find anything else, we have a parse error!
926 if (length $json->incr_text) {
927 die "parse error near ", $json->incr_text;
928 }
929
930 # else add more data
931 sysread $fh, my $buf, 65536
932 or die "read error: $!";
933 $json->incr_parse ($buf); # void context, so no parsing
934 }
935
936This is a complex example, but most of the complexity comes from the fact
937that we are trying to be correct (bear with me if I am wrong, I never ran
938the above example :).
939
940
319 941
320=head1 MAPPING 942=head1 MAPPING
321 943
322This section describes how JSON::XS maps Perl values to JSON values and 944This section describes how JSON::XS maps Perl values to JSON values and
323vice versa. These mappings are designed to "do the right thing" in most 945vice versa. These mappings are designed to "do the right thing" in most
324circumstances automatically, preserving round-tripping characteristics 946circumstances automatically, preserving round-tripping characteristics
325(what you put in comes out as something equivalent). 947(what you put in comes out as something equivalent).
326 948
327For the more enlightened: note that in the following descriptions, 949For the more enlightened: note that in the following descriptions,
328lowercase I<perl> refers to the Perl interpreter, while uppcercase I<Perl> 950lowercase I<perl> refers to the Perl interpreter, while uppercase I<Perl>
329refers to the abstract Perl language itself. 951refers to the abstract Perl language itself.
330 952
953
331=head2 JSON -> PERL 954=head2 JSON -> PERL
332 955
333=over 4 956=over 4
334 957
335=item object 958=item object
336 959
337A JSON object becomes a reference to a hash in Perl. No ordering of object 960A JSON object becomes a reference to a hash in Perl. No ordering of object
338keys is preserved (JSON does not preserver object key ordering itself). 961keys is preserved (JSON does not preserve object key ordering itself).
339 962
340=item array 963=item array
341 964
342A JSON array becomes a reference to an array in Perl. 965A JSON array becomes a reference to an array in Perl.
343 966
347are represented by the same codepoints in the Perl string, so no manual 970are represented by the same codepoints in the Perl string, so no manual
348decoding is necessary. 971decoding is necessary.
349 972
350=item number 973=item number
351 974
352A JSON number becomes either an integer or numeric (floating point) 975A JSON number becomes either an integer, numeric (floating point) or
353scalar in perl, depending on its range and any fractional parts. On the 976string scalar in perl, depending on its range and any fractional parts. On
354Perl level, there is no difference between those as Perl handles all the 977the Perl level, there is no difference between those as Perl handles all
355conversion details, but an integer may take slightly less memory and might 978the conversion details, but an integer may take slightly less memory and
356represent more values exactly than (floating point) numbers. 979might represent more values exactly than floating point numbers.
980
981If the number consists of digits only, JSON::XS will try to represent
982it as an integer value. If that fails, it will try to represent it as
983a numeric (floating point) value if that is possible without loss of
984precision. Otherwise it will preserve the number as a string value (in
985which case you lose roundtripping ability, as the JSON number will be
986re-encoded toa JSON string).
987
988Numbers containing a fractional or exponential part will always be
989represented as numeric (floating point) values, possibly at a loss of
990precision (in which case you might lose perfect roundtripping ability, but
991the JSON number will still be re-encoded as a JSON number).
357 992
358=item true, false 993=item true, false
359 994
360These JSON atoms become C<0>, C<1>, respectively. Information is lost in 995These JSON atoms become C<JSON::XS::true> and C<JSON::XS::false>,
361this process. Future versions might represent those values differently, 996respectively. They are overloaded to act almost exactly like the numbers
362but they will be guarenteed to act like these integers would normally in 997C<1> and C<0>. You can check whether a scalar is a JSON boolean by using
363Perl. 998the C<JSON::XS::is_bool> function.
364 999
365=item null 1000=item null
366 1001
367A JSON null atom becomes C<undef> in Perl. 1002A JSON null atom becomes C<undef> in Perl.
368 1003
369=back 1004=back
1005
370 1006
371=head2 PERL -> JSON 1007=head2 PERL -> JSON
372 1008
373The mapping from Perl to JSON is slightly more difficult, as Perl is a 1009The mapping from Perl to JSON is slightly more difficult, as Perl is a
374truly typeless language, so we can only guess which JSON type is meant by 1010truly typeless language, so we can only guess which JSON type is meant by
377=over 4 1013=over 4
378 1014
379=item hash references 1015=item hash references
380 1016
381Perl hash references become JSON objects. As there is no inherent ordering 1017Perl hash references become JSON objects. As there is no inherent ordering
382in hash keys, they will usually be encoded in a pseudo-random order that 1018in hash keys (or JSON objects), they will usually be encoded in a
383can change between runs of the same program but stays generally the same 1019pseudo-random order that can change between runs of the same program but
384within a single run of a program. JSON::XS can optionally sort the hash 1020stays generally the same within a single run of a program. JSON::XS can
385keys (determined by the I<canonical> flag), so the same datastructure 1021optionally sort the hash keys (determined by the I<canonical> flag), so
386will serialise to the same JSON text (given same settings and version of 1022the same datastructure will serialise to the same JSON text (given same
387JSON::XS), but this incurs a runtime overhead. 1023settings and version of JSON::XS), but this incurs a runtime overhead
1024and is only rarely useful, e.g. when you want to compare some JSON text
1025against another for equality.
388 1026
389=item array references 1027=item array references
390 1028
391Perl array references become JSON arrays. 1029Perl array references become JSON arrays.
392 1030
1031=item other references
1032
1033Other unblessed references are generally not allowed and will cause an
1034exception to be thrown, except for references to the integers C<0> and
1035C<1>, which get turned into C<false> and C<true> atoms in JSON. You can
1036also use C<JSON::XS::false> and C<JSON::XS::true> to improve readability.
1037
1038 encode_json [\0, JSON::XS::true] # yields [false,true]
1039
1040=item JSON::XS::true, JSON::XS::false
1041
1042These special values become JSON true and JSON false values,
1043respectively. You can also use C<\1> and C<\0> directly if you want.
1044
393=item blessed objects 1045=item blessed objects
394 1046
395Blessed objects are not allowed. JSON::XS currently tries to encode their 1047Blessed objects are not directly representable in JSON. See the
396underlying representation (hash- or arrayref), but this behaviour might 1048C<allow_blessed> and C<convert_blessed> methods on various options on
397change in future versions. 1049how to deal with this: basically, you can choose between throwing an
1050exception, encoding the reference as if it weren't blessed, or provide
1051your own serialiser method.
398 1052
399=item simple scalars 1053=item simple scalars
400 1054
401Simple Perl scalars (any scalar that is not a reference) are the most 1055Simple Perl scalars (any scalar that is not a reference) are the most
402difficult objects to encode: JSON::XS will encode undefined scalars as 1056difficult objects to encode: JSON::XS will encode undefined scalars as
403JSON null value, scalars that have last been used in a string context 1057JSON C<null> values, scalars that have last been used in a string context
404before encoding as JSON strings and anything else as number value: 1058before encoding as JSON strings, and anything else as number value:
405 1059
406 # dump as number 1060 # dump as number
407 to_json [2] # yields [2] 1061 encode_json [2] # yields [2]
408 to_json [-3.0e17] # yields [-3e+17] 1062 encode_json [-3.0e17] # yields [-3e+17]
409 my $value = 5; to_json [$value] # yields [5] 1063 my $value = 5; encode_json [$value] # yields [5]
410 1064
411 # used as string, so dump as string 1065 # used as string, so dump as string
412 print $value; 1066 print $value;
413 to_json [$value] # yields ["5"] 1067 encode_json [$value] # yields ["5"]
414 1068
415 # undef becomes null 1069 # undef becomes null
416 to_json [undef] # yields [null] 1070 encode_json [undef] # yields [null]
417 1071
418You can force the type to be a string by stringifying it: 1072You can force the type to be a JSON string by stringifying it:
419 1073
420 my $x = 3.1; # some variable containing a number 1074 my $x = 3.1; # some variable containing a number
421 "$x"; # stringified 1075 "$x"; # stringified
422 $x .= ""; # another, more awkward way to stringify 1076 $x .= ""; # another, more awkward way to stringify
423 print $x; # perl does it for you, too, quite often 1077 print $x; # perl does it for you, too, quite often
424 1078
425You can force the type to be a number by numifying it: 1079You can force the type to be a JSON number by numifying it:
426 1080
427 my $x = "3"; # some variable containing a string 1081 my $x = "3"; # some variable containing a string
428 $x += 0; # numify it, ensuring it will be dumped as a number 1082 $x += 0; # numify it, ensuring it will be dumped as a number
429 $x *= 1; # same thing, the choise is yours. 1083 $x *= 1; # same thing, the choice is yours.
430 1084
431You can not currently output JSON booleans or force the type in other, 1085You can not currently force the type in other, less obscure, ways. Tell me
432less obscure, ways. Tell me if you need this capability. 1086if you need this capability (but don't forget to explain why it's needed
433 1087:).
434=item circular data structures
435
436Those will be encoded until memory or stackspace runs out.
437 1088
438=back 1089=back
439 1090
440=head1 COMPARISON
441 1091
442As already mentioned, this module was created because none of the existing 1092=head1 ENCODING/CODESET FLAG NOTES
443JSON modules could be made to work correctly. First I will describe the 1093
444problems (or pleasures) I encountered with various existing JSON modules, 1094The interested reader might have seen a number of flags that signify
445followed by some benchmark values. JSON::XS was designed not to suffer 1095encodings or codesets - C<utf8>, C<latin1> and C<ascii>. There seems to be
446from any of these problems or limitations. 1096some confusion on what these do, so here is a short comparison:
1097
1098C<utf8> controls whether the JSON text created by C<encode> (and expected
1099by C<decode>) is UTF-8 encoded or not, while C<latin1> and C<ascii> only
1100control whether C<encode> escapes character values outside their respective
1101codeset range. Neither of these flags conflict with each other, although
1102some combinations make less sense than others.
1103
1104Care has been taken to make all flags symmetrical with respect to
1105C<encode> and C<decode>, that is, texts encoded with any combination of
1106these flag values will be correctly decoded when the same flags are used
1107- in general, if you use different flag settings while encoding vs. when
1108decoding you likely have a bug somewhere.
1109
1110Below comes a verbose discussion of these flags. Note that a "codeset" is
1111simply an abstract set of character-codepoint pairs, while an encoding
1112takes those codepoint numbers and I<encodes> them, in our case into
1113octets. Unicode is (among other things) a codeset, UTF-8 is an encoding,
1114and ISO-8859-1 (= latin 1) and ASCII are both codesets I<and> encodings at
1115the same time, which can be confusing.
447 1116
448=over 4 1117=over 4
449 1118
450=item JSON 1.07 1119=item C<utf8> flag disabled
451 1120
452Slow (but very portable, as it is written in pure Perl). 1121When C<utf8> is disabled (the default), then C<encode>/C<decode> generate
1122and expect Unicode strings, that is, characters with high ordinal Unicode
1123values (> 255) will be encoded as such characters, and likewise such
1124characters are decoded as-is, no canges to them will be done, except
1125"(re-)interpreting" them as Unicode codepoints or Unicode characters,
1126respectively (to Perl, these are the same thing in strings unless you do
1127funny/weird/dumb stuff).
453 1128
454Undocumented/buggy Unicode handling (how JSON handles unicode values is 1129This is useful when you want to do the encoding yourself (e.g. when you
455undocumented. One can get far by feeding it unicode strings and doing 1130want to have UTF-16 encoded JSON texts) or when some other layer does
456en-/decoding oneself, but unicode escapes are not working properly). 1131the encoding for you (for example, when printing to a terminal using a
1132filehandle that transparently encodes to UTF-8 you certainly do NOT want
1133to UTF-8 encode your data first and have Perl encode it another time).
457 1134
458No roundtripping (strings get clobbered if they look like numbers, e.g. 1135=item C<utf8> flag enabled
459the string C<2.0> will encode to C<2.0> instead of C<"2.0">, and that will
460decode into the number 2.
461 1136
462=item JSON::PC 0.01 1137If the C<utf8>-flag is enabled, C<encode>/C<decode> will encode all
1138characters using the corresponding UTF-8 multi-byte sequence, and will
1139expect your input strings to be encoded as UTF-8, that is, no "character"
1140of the input string must have any value > 255, as UTF-8 does not allow
1141that.
463 1142
464Very fast. 1143The C<utf8> flag therefore switches between two modes: disabled means you
1144will get a Unicode string in Perl, enabled means you get an UTF-8 encoded
1145octet/binary string in Perl.
465 1146
466Undocumented/buggy Unicode handling. 1147=item C<latin1> or C<ascii> flags enabled
467 1148
468No roundtripping. 1149With C<latin1> (or C<ascii>) enabled, C<encode> will escape characters
1150with ordinal values > 255 (> 127 with C<ascii>) and encode the remaining
1151characters as specified by the C<utf8> flag.
469 1152
470Has problems handling many Perl values (e.g. regex results and other magic 1153If C<utf8> is disabled, then the result is also correctly encoded in those
471values will make it croak). 1154character sets (as both are proper subsets of Unicode, meaning that a
1155Unicode string with all character values < 256 is the same thing as a
1156ISO-8859-1 string, and a Unicode string with all character values < 128 is
1157the same thing as an ASCII string in Perl).
472 1158
473Does not even generate valid JSON (C<{1,2}> gets converted to C<{1:2}> 1159If C<utf8> is enabled, you still get a correct UTF-8-encoded string,
474which is not a valid JSON text. 1160regardless of these flags, just some more characters will be escaped using
1161C<\uXXXX> then before.
475 1162
476Unmaintained (maintainer unresponsive for many months, bugs are not 1163Note that ISO-8859-1-I<encoded> strings are not compatible with UTF-8
477getting fixed). 1164encoding, while ASCII-encoded strings are. That is because the ISO-8859-1
1165encoding is NOT a subset of UTF-8 (despite the ISO-8859-1 I<codeset> being
1166a subset of Unicode), while ASCII is.
478 1167
479=item JSON::Syck 0.21 1168Surprisingly, C<decode> will ignore these flags and so treat all input
1169values as governed by the C<utf8> flag. If it is disabled, this allows you
1170to decode ISO-8859-1- and ASCII-encoded strings, as both strict subsets of
1171Unicode. If it is enabled, you can correctly decode UTF-8 encoded strings.
480 1172
481Very buggy (often crashes). 1173So neither C<latin1> nor C<ascii> are incompatible with the C<utf8> flag -
1174they only govern when the JSON output engine escapes a character or not.
482 1175
483Very inflexible (no human-readable format supported, format pretty much 1176The main use for C<latin1> is to relatively efficiently store binary data
484undocumented. I need at least a format for easy reading by humans and a 1177as JSON, at the expense of breaking compatibility with most JSON decoders.
485single-line compact format for use in a protocol, and preferably a way to
486generate ASCII-only JSON texts).
487 1178
488Completely broken (and confusingly documented) Unicode handling (unicode 1179The main use for C<ascii> is to force the output to not contain characters
489escapes are not working properly, you need to set ImplicitUnicode to 1180with values > 127, which means you can interpret the resulting string
490I<different> values on en- and decoding to get symmetric behaviour). 1181as UTF-8, ISO-8859-1, ASCII, KOI8-R or most about any character set and
491 11828-bit-encoding, and still get the same data structure back. This is useful
492No roundtripping (simple cases work, but this depends on wether the scalar 1183when your channel for JSON transfer is not 8-bit clean or the encoding
493value was used in a numeric context or not). 1184might be mangled in between (e.g. in mail), and works because ASCII is a
494 1185proper subset of most 8-bit and multibyte encodings in use in the world.
495Dumping hashes may skip hash values depending on iterator state.
496
497Unmaintained (maintainer unresponsive for many months, bugs are not
498getting fixed).
499
500Does not check input for validity (i.e. will accept non-JSON input and
501return "something" instead of raising an exception. This is a security
502issue: imagine two banks transfering money between each other using
503JSON. One bank might parse a given non-JSON request and deduct money,
504while the other might reject the transaction with a syntax error. While a
505good protocol will at least recover, that is extra unnecessary work and
506the transaction will still not succeed).
507
508=item JSON::DWIW 0.04
509
510Very fast. Very natural. Very nice.
511
512Undocumented unicode handling (but the best of the pack. Unicode escapes
513still don't get parsed properly).
514
515Very inflexible.
516
517No roundtripping.
518
519Does not generate valid JSON texts (key strings are often unquoted, empty keys
520result in nothing being output)
521
522Does not check input for validity.
523 1186
524=back 1187=back
1188
1189
1190=head2 JSON and ECMAscript
1191
1192JSON syntax is based on how literals are represented in javascript (the
1193not-standardised predecessor of ECMAscript) which is presumably why it is
1194called "JavaScript Object Notation".
1195
1196However, JSON is not a subset (and also not a superset of course) of
1197ECMAscript (the standard) or javascript (whatever browsers actually
1198implement).
1199
1200If you want to use javascript's C<eval> function to "parse" JSON, you
1201might run into parse errors for valid JSON texts, or the resulting data
1202structure might not be queryable:
1203
1204One of the problems is that U+2028 and U+2029 are valid characters inside
1205JSON strings, but are not allowed in ECMAscript string literals, so the
1206following Perl fragment will not output something that can be guaranteed
1207to be parsable by javascript's C<eval>:
1208
1209 use JSON::XS;
1210
1211 print encode_json [chr 0x2028];
1212
1213The right fix for this is to use a proper JSON parser in your javascript
1214programs, and not rely on C<eval>.
1215
1216If this is not an option, you can, as a stop-gap measure, simply encode to
1217ASCII-only JSON:
1218
1219 use JSON::XS;
1220
1221 print JSON::XS->new->ascii->encode ([chr 0x2028]);
1222
1223And if you are concerned about the size of the resulting JSON text, you
1224can run some regexes to only escape U+2028 and U+2029:
1225
1226 use JSON::XS;
1227
1228 my $json = JSON::XS->new->utf8->encode ([chr 0x2028]);
1229 $json =~ s/\xe2\x80\xa8/\\u2028/g; # escape U+2028
1230 $json =~ s/\xe2\x80\xa9/\\u2029/g; # escape U+2029
1231 print $json;
1232
1233This works because U+2028/U+2029 are not allowed outside of strings and
1234are not used for syntax, so replacing them unconditionally just works.
1235
1236Note, however, that fixing the broken JSON parser is better than working
1237around it in every other generator. The above regexes should work well in
1238other languages, as long as they operate on UTF-8. It is equally valid to
1239replace all occurences of U+2028/2029 directly by their \\u-escaped forms
1240in unicode texts, so they can simply be used to fix any parsers relying on
1241C<eval> by first applying the regexes on the encoded texts.
1242
1243Note also that the above only works for U+2028 and U+2029 and thus
1244only for fully ECMAscript-compliant parsers. Many existing javascript
1245implementations misparse other characters as well. Best rely on a good
1246JSON parser, such as Douglas Crockfords F<json2.js>, which escapes the
1247above and many more problematic characters properly before passing them
1248into C<eval>.
1249
1250Another problem is that some javascript implementations reserve
1251some property names for their own purposes (which probably makes
1252them non-ECMAscript-compliant). For example, Iceweasel reserves the
1253C<__proto__> property name for it's own purposes.
1254
1255If that is a problem, you could parse try to filter the resulting JSON
1256output for these property strings, e.g.:
1257
1258 $json =~ s/"__proto__"\s*:/"__proto__renamed":/g;
1259
1260This works because C<__proto__> is not valid outside of strings, so every
1261occurence of C<"__proto__"\s*:> must be a string used as property name.
1262
1263If you know of other incompatibilities, please let me know.
1264
1265
1266=head2 JSON and YAML
1267
1268You often hear that JSON is a subset of YAML. This is, however, a mass
1269hysteria(*) and very far from the truth (as of the time of this writing),
1270so let me state it clearly: I<in general, there is no way to configure
1271JSON::XS to output a data structure as valid YAML> that works in all
1272cases.
1273
1274If you really must use JSON::XS to generate YAML, you should use this
1275algorithm (subject to change in future versions):
1276
1277 my $to_yaml = JSON::XS->new->utf8->space_after (1);
1278 my $yaml = $to_yaml->encode ($ref) . "\n";
1279
1280This will I<usually> generate JSON texts that also parse as valid
1281YAML. Please note that YAML has hardcoded limits on (simple) object key
1282lengths that JSON doesn't have and also has different and incompatible
1283unicode handling, so you should make sure that your hash keys are
1284noticeably shorter than the 1024 "stream characters" YAML allows and that
1285you do not have characters with codepoint values outside the Unicode BMP
1286(basic multilingual page). YAML also does not allow C<\/> sequences in
1287strings (which JSON::XS does not I<currently> generate, but other JSON
1288generators might).
1289
1290There might be other incompatibilities that I am not aware of (or the YAML
1291specification has been changed yet again - it does so quite often). In
1292general you should not try to generate YAML with a JSON generator or vice
1293versa, or try to parse JSON with a YAML parser or vice versa: chances are
1294high that you will run into severe interoperability problems when you
1295least expect it.
1296
1297=over 4
1298
1299=item (*)
1300
1301I have been pressured multiple times by Brian Ingerson (one of the
1302authors of the YAML specification) to remove this paragraph, despite him
1303acknowledging that the actual incompatibilities exist. As I was personally
1304bitten by this "JSON is YAML" lie, I refused and said I will continue to
1305educate people about these issues, so others do not run into the same
1306problem again and again. After this, Brian called me a (quote)I<complete
1307and worthless idiot>(unquote).
1308
1309In my opinion, instead of pressuring and insulting people who actually
1310clarify issues with YAML and the wrong statements of some of its
1311proponents, I would kindly suggest reading the JSON spec (which is not
1312that difficult or long) and finally make YAML compatible to it, and
1313educating users about the changes, instead of spreading lies about the
1314real compatibility for many I<years> and trying to silence people who
1315point out that it isn't true.
1316
1317=back
1318
525 1319
526=head2 SPEED 1320=head2 SPEED
527 1321
528It seems that JSON::XS is surprisingly fast, as shown in the following 1322It seems that JSON::XS is surprisingly fast, as shown in the following
529tables. They have been generated with the help of the C<eg/bench> program 1323tables. They have been generated with the help of the C<eg/bench> program
530in the JSON::XS distribution, to make it easy to compare on your own 1324in the JSON::XS distribution, to make it easy to compare on your own
531system. 1325system.
532 1326
533First comes a comparison between various modules using a very short JSON 1327First comes a comparison between various modules using
534string: 1328a very short single-line JSON string (also available at
1329L<http://dist.schmorp.de/misc/json/short.json>).
535 1330
536 {"method": "handleMessage", "params": ["user1", "we were just talking"], "id": null} 1331 {"method": "handleMessage", "params": ["user1",
1332 "we were just talking"], "id": null, "array":[1,11,234,-5,1e5,1e7,
1333 true, false]}
537 1334
538It shows the number of encodes/decodes per second (JSON::XS uses the 1335It shows the number of encodes/decodes per second (JSON::XS uses
539functional interface, while JSON::XS/2 uses the OO interface with 1336the functional interface, while JSON::XS/2 uses the OO interface
540pretty-printing and hashkey sorting enabled). Higher is better: 1337with pretty-printing and hashkey sorting enabled, JSON::XS/3 enables
1338shrink). Higher is better:
541 1339
542 module | encode | decode | 1340 module | encode | decode |
543 -----------|------------|------------| 1341 -----------|------------|------------|
544 JSON | 11488.516 | 7823.035 | 1342 JSON 1.x | 4990.842 | 4088.813 |
545 JSON::DWIW | 94708.054 | 129094.260 | 1343 JSON::DWIW | 51653.990 | 71575.154 |
546 JSON::PC | 63884.157 | 128528.212 | 1344 JSON::PC | 65948.176 | 74631.744 |
547 JSON::Syck | 34898.677 | 42096.911 | 1345 JSON::PP | 8931.652 | 3817.168 |
548 JSON::XS | 654027.064 | 396423.669 | 1346 JSON::Syck | 24877.248 | 27776.848 |
549 JSON::XS/2 | 371564.190 | 371725.613 | 1347 JSON::XS | 388361.481 | 227951.304 |
1348 JSON::XS/2 | 227951.304 | 218453.333 |
1349 JSON::XS/3 | 338250.323 | 218453.333 |
1350 Storable | 16500.016 | 135300.129 |
550 -----------+------------+------------+ 1351 -----------+------------+------------+
551 1352
552That is, JSON::XS is more than six times faster than JSON::DWIW on 1353That is, JSON::XS is about five times faster than JSON::DWIW on encoding,
553encoding, more than three times faster on decoding, and about thirty times 1354about three times faster on decoding, and over forty times faster
554faster than JSON, even with pretty-printing and key sorting. 1355than JSON, even with pretty-printing and key sorting. It also compares
1356favourably to Storable for small amounts of data.
555 1357
556Using a longer test string (roughly 18KB, generated from Yahoo! Locals 1358Using a longer test string (roughly 18KB, generated from Yahoo! Locals
557search API (http://nanoref.com/yahooapis/mgPdGg): 1359search API (L<http://dist.schmorp.de/misc/json/long.json>).
558 1360
559 module | encode | decode | 1361 module | encode | decode |
560 -----------|------------|------------| 1362 -----------|------------|------------|
561 JSON | 273.023 | 44.674 | 1363 JSON 1.x | 55.260 | 34.971 |
562 JSON::DWIW | 1089.383 | 1145.704 | 1364 JSON::DWIW | 825.228 | 1082.513 |
563 JSON::PC | 3097.419 | 2393.921 | 1365 JSON::PC | 3571.444 | 2394.829 |
564 JSON::Syck | 514.060 | 843.053 | 1366 JSON::PP | 210.987 | 32.574 |
565 JSON::XS | 6479.668 | 3636.364 | 1367 JSON::Syck | 552.551 | 787.544 |
566 JSON::XS/2 | 3774.221 | 3599.124 | 1368 JSON::XS | 5780.463 | 4854.519 |
1369 JSON::XS/2 | 3869.998 | 4798.975 |
1370 JSON::XS/3 | 5862.880 | 4798.975 |
1371 Storable | 4445.002 | 5235.027 |
567 -----------+------------+------------+ 1372 -----------+------------+------------+
568 1373
569Again, JSON::XS leads by far. 1374Again, JSON::XS leads by far (except for Storable which non-surprisingly
1375decodes faster).
570 1376
571On large strings containing lots of high unicode characters, some modules 1377On large strings containing lots of high Unicode characters, some modules
572(such as JSON::PC) seem to decode faster than JSON::XS, but the result 1378(such as JSON::PC) seem to decode faster than JSON::XS, but the result
573will be broken due to missing (or wrong) unicode handling. Others refuse 1379will be broken due to missing (or wrong) Unicode handling. Others refuse
574to decode or encode properly, so it was impossible to prepare a fair 1380to decode or encode properly, so it was impossible to prepare a fair
575comparison table for that case. 1381comparison table for that case.
576 1382
577=head1 RESOURCE LIMITS
578 1383
579JSON::XS does not impose any limits on the size of JSON texts or Perl 1384=head1 SECURITY CONSIDERATIONS
580values they represent - if your machine can handle it, JSON::XS will 1385
581encode or decode it. Future versions might optionally impose structure 1386When you are using JSON in a protocol, talking to untrusted potentially
582depth and memory use resource limits. 1387hostile creatures requires relatively few measures.
1388
1389First of all, your JSON decoder should be secure, that is, should not have
1390any buffer overflows. Obviously, this module should ensure that and I am
1391trying hard on making that true, but you never know.
1392
1393Second, you need to avoid resource-starving attacks. That means you should
1394limit the size of JSON texts you accept, or make sure then when your
1395resources run out, that's just fine (e.g. by using a separate process that
1396can crash safely). The size of a JSON text in octets or characters is
1397usually a good indication of the size of the resources required to decode
1398it into a Perl structure. While JSON::XS can check the size of the JSON
1399text, it might be too late when you already have it in memory, so you
1400might want to check the size before you accept the string.
1401
1402Third, JSON::XS recurses using the C stack when decoding objects and
1403arrays. The C stack is a limited resource: for instance, on my amd64
1404machine with 8MB of stack size I can decode around 180k nested arrays but
1405only 14k nested JSON objects (due to perl itself recursing deeply on croak
1406to free the temporary). If that is exceeded, the program crashes. To be
1407conservative, the default nesting limit is set to 512. If your process
1408has a smaller stack, you should adjust this setting accordingly with the
1409C<max_depth> method.
1410
1411Something else could bomb you, too, that I forgot to think of. In that
1412case, you get to keep the pieces. I am always open for hints, though...
1413
1414Also keep in mind that JSON::XS might leak contents of your Perl data
1415structures in its error messages, so when you serialise sensitive
1416information you might want to make sure that exceptions thrown by JSON::XS
1417will not end up in front of untrusted eyes.
1418
1419If you are using JSON::XS to return packets to consumption
1420by JavaScript scripts in a browser you should have a look at
1421L<http://jpsykes.com/47/practical-csrf-and-json-security> to see whether
1422you are vulnerable to some common attack vectors (which really are browser
1423design bugs, but it is still you who will have to deal with it, as major
1424browser developers care only for features, not about getting security
1425right).
1426
1427
1428=head1 THREADS
1429
1430This module is I<not> guaranteed to be thread safe and there are no
1431plans to change this until Perl gets thread support (as opposed to the
1432horribly slow so-called "threads" which are simply slow and bloated
1433process simulations - use fork, it's I<much> faster, cheaper, better).
1434
1435(It might actually work, but you have been warned).
1436
583 1437
584=head1 BUGS 1438=head1 BUGS
585 1439
586While the goal of this module is to be correct, that unfortunately does 1440While the goal of this module is to be correct, that unfortunately does
587not mean its bug-free, only that I think its design is bug-free. It is 1441not mean it's bug-free, only that I think its design is bug-free. If you
588still very young and not well-tested. If you keep reporting bugs they will 1442keep reporting bugs they will be fixed swiftly, though.
589be fixed swiftly, though. 1443
1444Please refrain from using rt.cpan.org or any other bug reporting
1445service. I put the contact address into my modules for a reason.
590 1446
591=cut 1447=cut
592 1448
1449our $true = do { bless \(my $dummy = 1), "JSON::XS::Boolean" };
1450our $false = do { bless \(my $dummy = 0), "JSON::XS::Boolean" };
1451
1452sub true() { $true }
1453sub false() { $false }
1454
1455sub is_bool($) {
1456 UNIVERSAL::isa $_[0], "JSON::XS::Boolean"
1457# or UNIVERSAL::isa $_[0], "JSON::Literal"
1458}
1459
1460XSLoader::load "JSON::XS", $VERSION;
1461
1462package JSON::XS::Boolean;
1463
1464use overload
1465 "0+" => sub { ${$_[0]} },
1466 "++" => sub { $_[0] = ${$_[0]} + 1 },
1467 "--" => sub { $_[0] = ${$_[0]} - 1 },
1468 fallback => 1;
1469
5931; 14701;
1471
1472=head1 SEE ALSO
1473
1474The F<json_xs> command line utility for quick experiments.
594 1475
595=head1 AUTHOR 1476=head1 AUTHOR
596 1477
597 Marc Lehmann <schmorp@schmorp.de> 1478 Marc Lehmann <schmorp@schmorp.de>
598 http://home.schmorp.de/ 1479 http://home.schmorp.de/

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines