1NAME 2 JSON::XS - JSON serialising/deserialising, done correctly and fast 3 4 JSON::XS - 正しくて高速な JSON シリアライザ/デシリアライザ 5 (http://fleur.hio.jp/perldoc/mix/lib/JSON/XS.html) 6 7SYNOPSIS 8 use JSON::XS; 9 10 # exported functions, they croak on error 11 # and expect/generate UTF-8 12 13 $utf8_encoded_json_text = encode_json $perl_hash_or_arrayref; 14 $perl_hash_or_arrayref = decode_json $utf8_encoded_json_text; 15 16 # OO-interface 17 18 $coder = JSON::XS->new->ascii->pretty->allow_nonref; 19 $pretty_printed_unencoded = $coder->encode ($perl_scalar); 20 $perl_scalar = $coder->decode ($unicode_json_text); 21 22 # Note that JSON version 2.0 and above will automatically use JSON::XS 23 # if available, at virtually no speed overhead either, so you should 24 # be able to just: 25 26 use JSON; 27 28 # and do the same things, except that you have a pure-perl fallback now. 29 30DESCRIPTION 31 This module converts Perl data structures to JSON and vice versa. Its 32 primary goal is to be *correct* and its secondary goal is to be *fast*. 33 To reach the latter goal it was written in C. 34 35 See MAPPING, below, on how JSON::XS maps perl values to JSON values and 36 vice versa. 37 38 FEATURES 39 * correct Unicode handling 40 41 This module knows how to handle Unicode, documents how and when it 42 does so, and even documents what "correct" means. 43 44 * round-trip integrity 45 46 When you serialise a perl data structure using only data types 47 supported by JSON and Perl, the deserialised data structure is 48 identical on the Perl level. (e.g. the string "2.0" doesn't suddenly 49 become "2" just because it looks like a number). There *are* minor 50 exceptions to this, read the MAPPING section below to learn about 51 those. 52 53 * strict checking of JSON correctness 54 55 There is no guessing, no generating of illegal JSON texts by 56 default, and only JSON is accepted as input by default (the latter 57 is a security feature). 58 59 * fast 60 61 Compared to other JSON modules and other serialisers such as 62 Storable, this module usually compares favourably in terms of speed, 63 too. 64 65 * simple to use 66 67 This module has both a simple functional interface as well as an 68 object oriented interface. 69 70 * reasonably versatile output formats 71 72 You can choose between the most compact guaranteed-single-line 73 format possible (nice for simple line-based protocols), a pure-ASCII 74 format (for when your transport is not 8-bit clean, still supports 75 the whole Unicode range), or a pretty-printed format (for when you 76 want to read that stuff). Or you can combine those features in 77 whatever way you like. 78 79FUNCTIONAL INTERFACE 80 The following convenience methods are provided by this module. They are 81 exported by default: 82 83 $json_text = encode_json $perl_scalar 84 Converts the given Perl data structure to a UTF-8 encoded, binary 85 string (that is, the string contains octets only). Croaks on error. 86 87 This function call is functionally identical to: 88 89 $json_text = JSON::XS->new->utf8->encode ($perl_scalar) 90 91 Except being faster. 92 93 $perl_scalar = decode_json $json_text 94 The opposite of "encode_json": expects a UTF-8 (binary) string and 95 tries to parse that as a UTF-8 encoded JSON text, returning the 96 resulting reference. Croaks on error. 97 98 This function call is functionally identical to: 99 100 $perl_scalar = JSON::XS->new->utf8->decode ($json_text) 101 102 Except being faster. 103 104A FEW NOTES ON UNICODE AND PERL 105 Since this often leads to confusion, here are a few very clear words on 106 how Unicode works in Perl, modulo bugs. 107 108 1. Perl strings can store characters with ordinal values > 255. 109 This enables you to store Unicode characters as single characters in 110 a Perl string - very natural. 111 112 2. Perl does *not* associate an encoding with your strings. 113 ... until you force it to, e.g. when matching it against a regex, or 114 printing the scalar to a file, in which case Perl either interprets 115 your string as locale-encoded text, octets/binary, or as Unicode, 116 depending on various settings. In no case is an encoding stored 117 together with your data, it is *use* that decides encoding, not any 118 magical meta data. 119 120 3. The internal utf-8 flag has no meaning with regards to the encoding 121 of your string. 122 Just ignore that flag unless you debug a Perl bug, a module written 123 in XS or want to dive into the internals of perl. Otherwise it will 124 only confuse you, as, despite the name, it says nothing about how 125 your string is encoded. You can have Unicode strings with that flag 126 set, with that flag clear, and you can have binary data with that 127 flag set and that flag clear. Other possibilities exist, too. 128 129 If you didn't know about that flag, just the better, pretend it 130 doesn't exist. 131 132 4. A "Unicode String" is simply a string where each character can be 133 validly interpreted as a Unicode code point. 134 If you have UTF-8 encoded data, it is no longer a Unicode string, 135 but a Unicode string encoded in UTF-8, giving you a binary string. 136 137 5. A string containing "high" (> 255) character values is *not* a UTF-8 138 string. 139 It's a fact. Learn to live with it. 140 141 I hope this helps :) 142 143OBJECT-ORIENTED INTERFACE 144 The object oriented interface lets you configure your own encoding or 145 decoding style, within the limits of supported formats. 146 147 $json = new JSON::XS 148 Creates a new JSON::XS object that can be used to de/encode JSON 149 strings. All boolean flags described below are by default *disabled* 150 (with the exception of "allow_nonref", which defaults to *enabled* 151 since version 4.0). 152 153 The mutators for flags all return the JSON object again and thus 154 calls can be chained: 155 156 my $json = JSON::XS->new->utf8->space_after->encode ({a => [1,2]}) 157 => {"a": [1, 2]} 158 159 $json = $json->ascii ([$enable]) 160 $enabled = $json->get_ascii 161 If $enable is true (or missing), then the "encode" method will not 162 generate characters outside the code range 0..127 (which is ASCII). 163 Any Unicode characters outside that range will be escaped using 164 either a single \uXXXX (BMP characters) or a double \uHHHH\uLLLLL 165 escape sequence, as per RFC4627. The resulting encoded JSON text can 166 be treated as a native Unicode string, an ascii-encoded, 167 latin1-encoded or UTF-8 encoded string, or any other superset of 168 ASCII. 169 170 If $enable is false, then the "encode" method will not escape 171 Unicode characters unless required by the JSON syntax or other 172 flags. This results in a faster and more compact format. 173 174 See also the section *ENCODING/CODESET FLAG NOTES* later in this 175 document. 176 177 The main use for this flag is to produce JSON texts that can be 178 transmitted over a 7-bit channel, as the encoded JSON texts will not 179 contain any 8 bit characters. 180 181 JSON::XS->new->ascii (1)->encode ([chr 0x10401]) 182 => ["\ud801\udc01"] 183 184 $json = $json->latin1 ([$enable]) 185 $enabled = $json->get_latin1 186 If $enable is true (or missing), then the "encode" method will 187 encode the resulting JSON text as latin1 (or iso-8859-1), escaping 188 any characters outside the code range 0..255. The resulting string 189 can be treated as a latin1-encoded JSON text or a native Unicode 190 string. The "decode" method will not be affected in any way by this 191 flag, as "decode" by default expects Unicode, which is a strict 192 superset of latin1. 193 194 If $enable is false, then the "encode" method will not escape 195 Unicode characters unless required by the JSON syntax or other 196 flags. 197 198 See also the section *ENCODING/CODESET FLAG NOTES* later in this 199 document. 200 201 The main use for this flag is efficiently encoding binary data as 202 JSON text, as most octets will not be escaped, resulting in a 203 smaller encoded size. The disadvantage is that the resulting JSON 204 text is encoded in latin1 (and must correctly be treated as such 205 when storing and transferring), a rare encoding for JSON. It is 206 therefore most useful when you want to store data structures known 207 to contain binary data efficiently in files or databases, not when 208 talking to other JSON encoders/decoders. 209 210 JSON::XS->new->latin1->encode (["\x{89}\x{abc}"] 211 => ["\x{89}\\u0abc"] # (perl syntax, U+abc escaped, U+89 not) 212 213 $json = $json->utf8 ([$enable]) 214 $enabled = $json->get_utf8 215 If $enable is true (or missing), then the "encode" method will 216 encode the JSON result into UTF-8, as required by many protocols, 217 while the "decode" method expects to be handed a UTF-8-encoded 218 string. Please note that UTF-8-encoded strings do not contain any 219 characters outside the range 0..255, they are thus useful for 220 bytewise/binary I/O. In future versions, enabling this option might 221 enable autodetection of the UTF-16 and UTF-32 encoding families, as 222 described in RFC4627. 223 224 If $enable is false, then the "encode" method will return the JSON 225 string as a (non-encoded) Unicode string, while "decode" expects 226 thus a Unicode string. Any decoding or encoding (e.g. to UTF-8 or 227 UTF-16) needs to be done yourself, e.g. using the Encode module. 228 229 See also the section *ENCODING/CODESET FLAG NOTES* later in this 230 document. 231 232 Example, output UTF-16BE-encoded JSON: 233 234 use Encode; 235 $jsontext = encode "UTF-16BE", JSON::XS->new->encode ($object); 236 237 Example, decode UTF-32LE-encoded JSON: 238 239 use Encode; 240 $object = JSON::XS->new->decode (decode "UTF-32LE", $jsontext); 241 242 $json = $json->pretty ([$enable]) 243 This enables (or disables) all of the "indent", "space_before" and 244 "space_after" (and in the future possibly more) flags in one call to 245 generate the most readable (or most compact) form possible. 246 247 Example, pretty-print some simple structure: 248 249 my $json = JSON::XS->new->pretty(1)->encode ({a => [1,2]}) 250 => 251 { 252 "a" : [ 253 1, 254 2 255 ] 256 } 257 258 $json = $json->indent ([$enable]) 259 $enabled = $json->get_indent 260 If $enable is true (or missing), then the "encode" method will use a 261 multiline format as output, putting every array member or 262 object/hash key-value pair into its own line, indenting them 263 properly. 264 265 If $enable is false, no newlines or indenting will be produced, and 266 the resulting JSON text is guaranteed not to contain any "newlines". 267 268 This setting has no effect when decoding JSON texts. 269 270 $json = $json->space_before ([$enable]) 271 $enabled = $json->get_space_before 272 If $enable is true (or missing), then the "encode" method will add 273 an extra optional space before the ":" separating keys from values 274 in JSON objects. 275 276 If $enable is false, then the "encode" method will not add any extra 277 space at those places. 278 279 This setting has no effect when decoding JSON texts. You will also 280 most likely combine this setting with "space_after". 281 282 Example, space_before enabled, space_after and indent disabled: 283 284 {"key" :"value"} 285 286 $json = $json->space_after ([$enable]) 287 $enabled = $json->get_space_after 288 If $enable is true (or missing), then the "encode" method will add 289 an extra optional space after the ":" separating keys from values in 290 JSON objects and extra whitespace after the "," separating key-value 291 pairs and array members. 292 293 If $enable is false, then the "encode" method will not add any extra 294 space at those places. 295 296 This setting has no effect when decoding JSON texts. 297 298 Example, space_before and indent disabled, space_after enabled: 299 300 {"key": "value"} 301 302 $json = $json->relaxed ([$enable]) 303 $enabled = $json->get_relaxed 304 If $enable is true (or missing), then "decode" will accept some 305 extensions to normal JSON syntax (see below). "encode" will not be 306 affected in any way. *Be aware that this option makes you accept 307 invalid JSON texts as if they were valid!*. I suggest only to use 308 this option to parse application-specific files written by humans 309 (configuration files, resource files etc.) 310 311 If $enable is false (the default), then "decode" will only accept 312 valid JSON texts. 313 314 Currently accepted extensions are: 315 316 * list items can have an end-comma 317 318 JSON *separates* array elements and key-value pairs with commas. 319 This can be annoying if you write JSON texts manually and want 320 to be able to quickly append elements, so this extension accepts 321 comma at the end of such items not just between them: 322 323 [ 324 1, 325 2, <- this comma not normally allowed 326 ] 327 { 328 "k1": "v1", 329 "k2": "v2", <- this comma not normally allowed 330 } 331 332 * shell-style '#'-comments 333 334 Whenever JSON allows whitespace, shell-style comments are 335 additionally allowed. They are terminated by the first 336 carriage-return or line-feed character, after which more 337 white-space and comments are allowed. 338 339 [ 340 1, # this comment not allowed in JSON 341 # neither this one... 342 ] 343 344 * literal ASCII TAB characters in strings 345 346 Literal ASCII TAB characters are now allowed in strings (and 347 treated as "\t"). 348 349 [ 350 "Hello\tWorld", 351 "Hello<TAB>World", # literal <TAB> would not normally be allowed 352 ] 353 354 $json = $json->canonical ([$enable]) 355 $enabled = $json->get_canonical 356 If $enable is true (or missing), then the "encode" method will 357 output JSON objects by sorting their keys. This is adding a 358 comparatively high overhead. 359 360 If $enable is false, then the "encode" method will output key-value 361 pairs in the order Perl stores them (which will likely change 362 between runs of the same script, and can change even within the same 363 run from 5.18 onwards). 364 365 This option is useful if you want the same data structure to be 366 encoded as the same JSON text (given the same overall settings). If 367 it is disabled, the same hash might be encoded differently even if 368 contains the same data, as key-value pairs have no inherent ordering 369 in Perl. 370 371 This setting has no effect when decoding JSON texts. 372 373 This setting has currently no effect on tied hashes. 374 375 $json = $json->allow_nonref ([$enable]) 376 $enabled = $json->get_allow_nonref 377 Unlike other boolean options, this opotion is enabled by default 378 beginning with version 4.0. See "SECURITY CONSIDERATIONS" for the 379 gory details. 380 381 If $enable is true (or missing), then the "encode" method can 382 convert a non-reference into its corresponding string, number or 383 null JSON value, which is an extension to RFC4627. Likewise, 384 "decode" will accept those JSON values instead of croaking. 385 386 If $enable is false, then the "encode" method will croak if it isn't 387 passed an arrayref or hashref, as JSON texts must either be an 388 object or array. Likewise, "decode" will croak if given something 389 that is not a JSON object or array. 390 391 Example, encode a Perl scalar as JSON value without enabled 392 "allow_nonref", resulting in an error: 393 394 JSON::XS->new->allow_nonref (0)->encode ("Hello, World!") 395 => hash- or arrayref expected... 396 397 $json = $json->allow_unknown ([$enable]) 398 $enabled = $json->get_allow_unknown 399 If $enable is true (or missing), then "encode" will *not* throw an 400 exception when it encounters values it cannot represent in JSON (for 401 example, filehandles) but instead will encode a JSON "null" value. 402 Note that blessed objects are not included here and are handled 403 separately by c<allow_nonref>. 404 405 If $enable is false (the default), then "encode" will throw an 406 exception when it encounters anything it cannot encode as JSON. 407 408 This option does not affect "decode" in any way, and it is 409 recommended to leave it off unless you know your communications 410 partner. 411 412 $json = $json->allow_blessed ([$enable]) 413 $enabled = $json->get_allow_blessed 414 See "OBJECT SERIALISATION" for details. 415 416 If $enable is true (or missing), then the "encode" method will not 417 barf when it encounters a blessed reference that it cannot convert 418 otherwise. Instead, a JSON "null" value is encoded instead of the 419 object. 420 421 If $enable is false (the default), then "encode" will throw an 422 exception when it encounters a blessed object that it cannot convert 423 otherwise. 424 425 This setting has no effect on "decode". 426 427 $json = $json->convert_blessed ([$enable]) 428 $enabled = $json->get_convert_blessed 429 See "OBJECT SERIALISATION" for details. 430 431 If $enable is true (or missing), then "encode", upon encountering a 432 blessed object, will check for the availability of the "TO_JSON" 433 method on the object's class. If found, it will be called in scalar 434 context and the resulting scalar will be encoded instead of the 435 object. 436 437 The "TO_JSON" method may safely call die if it wants. If "TO_JSON" 438 returns other blessed objects, those will be handled in the same 439 way. "TO_JSON" must take care of not causing an endless recursion 440 cycle (== crash) in this case. The name of "TO_JSON" was chosen 441 because other methods called by the Perl core (== not by the user of 442 the object) are usually in upper case letters and to avoid 443 collisions with any "to_json" function or method. 444 445 If $enable is false (the default), then "encode" will not consider 446 this type of conversion. 447 448 This setting has no effect on "decode". 449 450 $json = $json->allow_tags ([$enable]) 451 $enabled = $json->get_allow_tags 452 See "OBJECT SERIALISATION" for details. 453 454 If $enable is true (or missing), then "encode", upon encountering a 455 blessed object, will check for the availability of the "FREEZE" 456 method on the object's class. If found, it will be used to serialise 457 the object into a nonstandard tagged JSON value (that JSON decoders 458 cannot decode). 459 460 It also causes "decode" to parse such tagged JSON values and 461 deserialise them via a call to the "THAW" method. 462 463 If $enable is false (the default), then "encode" will not consider 464 this type of conversion, and tagged JSON values will cause a parse 465 error in "decode", as if tags were not part of the grammar. 466 467 $json->boolean_values ([$false, $true]) 468 ($false, $true) = $json->get_boolean_values 469 By default, JSON booleans will be decoded as overloaded 470 $Types::Serialiser::false and $Types::Serialiser::true objects. 471 472 With this method you can specify your own boolean values for 473 decoding - on decode, JSON "false" will be decoded as a copy of 474 $false, and JSON "true" will be decoded as $true ("copy" here is the 475 same thing as assigning a value to another variable, i.e. "$copy = 476 $false"). 477 478 Calling this method without any arguments will reset the booleans to 479 their default values. 480 481 "get_boolean_values" will return both $false and $true values, or 482 the empty list when they are set to the default. 483 484 $json = $json->filter_json_object ([$coderef->($hashref)]) 485 When $coderef is specified, it will be called from "decode" each 486 time it decodes a JSON object. The only argument is a reference to 487 the newly-created hash. If the code reference returns a single 488 scalar (which need not be a reference), this value (or rather a copy 489 of it) is inserted into the deserialised data structure. If it 490 returns an empty list (NOTE: *not* "undef", which is a valid 491 scalar), the original deserialised hash will be inserted. This 492 setting can slow down decoding considerably. 493 494 When $coderef is omitted or undefined, any existing callback will be 495 removed and "decode" will not change the deserialised hash in any 496 way. 497 498 Example, convert all JSON objects into the integer 5: 499 500 my $js = JSON::XS->new->filter_json_object (sub { 5 }); 501 # returns [5] 502 $js->decode ('[{}]') 503 # throw an exception because allow_nonref is not enabled 504 # so a lone 5 is not allowed. 505 $js->decode ('{"a":1, "b":2}'); 506 507 $json = $json->filter_json_single_key_object ($key [=> 508 $coderef->($value)]) 509 Works remotely similar to "filter_json_object", but is only called 510 for JSON objects having a single key named $key. 511 512 This $coderef is called before the one specified via 513 "filter_json_object", if any. It gets passed the single value in the 514 JSON object. If it returns a single value, it will be inserted into 515 the data structure. If it returns nothing (not even "undef" but the 516 empty list), the callback from "filter_json_object" will be called 517 next, as if no single-key callback were specified. 518 519 If $coderef is omitted or undefined, the corresponding callback will 520 be disabled. There can only ever be one callback for a given key. 521 522 As this callback gets called less often then the 523 "filter_json_object" one, decoding speed will not usually suffer as 524 much. Therefore, single-key objects make excellent targets to 525 serialise Perl objects into, especially as single-key JSON objects 526 are as close to the type-tagged value concept as JSON gets (it's 527 basically an ID/VALUE tuple). Of course, JSON does not support this 528 in any way, so you need to make sure your data never looks like a 529 serialised Perl hash. 530 531 Typical names for the single object key are "__class_whatever__", or 532 "$__dollars_are_rarely_used__$" or "}ugly_brace_placement", or even 533 things like "__class_md5sum(classname)__", to reduce the risk of 534 clashing with real hashes. 535 536 Example, decode JSON objects of the form "{ "__widget__" => <id> }" 537 into the corresponding $WIDGET{<id>} object: 538 539 # return whatever is in $WIDGET{5}: 540 JSON::XS 541 ->new 542 ->filter_json_single_key_object (__widget__ => sub { 543 $WIDGET{ $_[0] } 544 }) 545 ->decode ('{"__widget__": 5') 546 547 # this can be used with a TO_JSON method in some "widget" class 548 # for serialisation to json: 549 sub WidgetBase::TO_JSON { 550 my ($self) = @_; 551 552 unless ($self->{id}) { 553 $self->{id} = ..get..some..id..; 554 $WIDGET{$self->{id}} = $self; 555 } 556 557 { __widget__ => $self->{id} } 558 } 559 560 $json = $json->shrink ([$enable]) 561 $enabled = $json->get_shrink 562 Perl usually over-allocates memory a bit when allocating space for 563 strings. This flag optionally resizes strings generated by either 564 "encode" or "decode" to their minimum size possible. This can save 565 memory when your JSON texts are either very very long or you have 566 many short strings. It will also try to downgrade any strings to 567 octet-form if possible: perl stores strings internally either in an 568 encoding called UTF-X or in octet-form. The latter cannot store 569 everything but uses less space in general (and some buggy Perl or C 570 code might even rely on that internal representation being used). 571 572 The actual definition of what shrink does might change in future 573 versions, but it will always try to save space at the expense of 574 time. 575 576 If $enable is true (or missing), the string returned by "encode" 577 will be shrunk-to-fit, while all strings generated by "decode" will 578 also be shrunk-to-fit. 579 580 If $enable is false, then the normal perl allocation algorithms are 581 used. If you work with your data, then this is likely to be faster. 582 583 In the future, this setting might control other things, such as 584 converting strings that look like integers or floats into integers 585 or floats internally (there is no difference on the Perl level), 586 saving space. 587 588 $json = $json->max_depth ([$maximum_nesting_depth]) 589 $max_depth = $json->get_max_depth 590 Sets the maximum nesting level (default 512) accepted while encoding 591 or decoding. If a higher nesting level is detected in JSON text or a 592 Perl data structure, then the encoder and decoder will stop and 593 croak at that point. 594 595 Nesting level is defined by number of hash- or arrayrefs that the 596 encoder needs to traverse to reach a given point or the number of 597 "{" or "[" characters without their matching closing parenthesis 598 crossed to reach a given character in a string. 599 600 Setting the maximum depth to one disallows any nesting, so that 601 ensures that the object is only a single hash/object or array. 602 603 If no argument is given, the highest possible setting will be used, 604 which is rarely useful. 605 606 Note that nesting is implemented by recursion in C. The default 607 value has been chosen to be as large as typical operating systems 608 allow without crashing. 609 610 See SECURITY CONSIDERATIONS, below, for more info on why this is 611 useful. 612 613 $json = $json->max_size ([$maximum_string_size]) 614 $max_size = $json->get_max_size 615 Set the maximum length a JSON text may have (in bytes) where 616 decoding is being attempted. The default is 0, meaning no limit. 617 When "decode" is called on a string that is longer then this many 618 bytes, it will not attempt to decode the string but throw an 619 exception. This setting has no effect on "encode" (yet). 620 621 If no argument is given, the limit check will be deactivated (same 622 as when 0 is specified). 623 624 See SECURITY CONSIDERATIONS, below, for more info on why this is 625 useful. 626 627 $json_text = $json->encode ($perl_scalar) 628 Converts the given Perl value or data structure to its JSON 629 representation. Croaks on error. 630 631 $perl_scalar = $json->decode ($json_text) 632 The opposite of "encode": expects a JSON text and tries to parse it, 633 returning the resulting simple scalar or reference. Croaks on error. 634 635 ($perl_scalar, $characters) = $json->decode_prefix ($json_text) 636 This works like the "decode" method, but instead of raising an 637 exception when there is trailing garbage after the first JSON 638 object, it will silently stop parsing there and return the number of 639 characters consumed so far. 640 641 This is useful if your JSON texts are not delimited by an outer 642 protocol and you need to know where the JSON text ends. 643 644 JSON::XS->new->decode_prefix ("[1] the tail") 645 => ([1], 3) 646 647INCREMENTAL PARSING 648 In some cases, there is the need for incremental parsing of JSON texts. 649 While this module always has to keep both JSON text and resulting Perl 650 data structure in memory at one time, it does allow you to parse a JSON 651 stream incrementally. It does so by accumulating text until it has a 652 full JSON object, which it then can decode. This process is similar to 653 using "decode_prefix" to see if a full JSON object is available, but is 654 much more efficient (and can be implemented with a minimum of method 655 calls). 656 657 JSON::XS will only attempt to parse the JSON text once it is sure it has 658 enough text to get a decisive result, using a very simple but truly 659 incremental parser. This means that it sometimes won't stop as early as 660 the full parser, for example, it doesn't detect mismatched parentheses. 661 The only thing it guarantees is that it starts decoding as soon as a 662 syntactically valid JSON text has been seen. This means you need to set 663 resource limits (e.g. "max_size") to ensure the parser will stop parsing 664 in the presence if syntax errors. 665 666 The following methods implement this incremental parser. 667 668 [void, scalar or list context] = $json->incr_parse ([$string]) 669 This is the central parsing function. It can both append new text 670 and extract objects from the stream accumulated so far (both of 671 these functions are optional). 672 673 If $string is given, then this string is appended to the already 674 existing JSON fragment stored in the $json object. 675 676 After that, if the function is called in void context, it will 677 simply return without doing anything further. This can be used to 678 add more text in as many chunks as you want. 679 680 If the method is called in scalar context, then it will try to 681 extract exactly *one* JSON object. If that is successful, it will 682 return this object, otherwise it will return "undef". If there is a 683 parse error, this method will croak just as "decode" would do (one 684 can then use "incr_skip" to skip the erroneous part). This is the 685 most common way of using the method. 686 687 And finally, in list context, it will try to extract as many objects 688 from the stream as it can find and return them, or the empty list 689 otherwise. For this to work, there must be no separators (other than 690 whitespace) between the JSON objects or arrays, instead they must be 691 concatenated back-to-back. If an error occurs, an exception will be 692 raised as in the scalar context case. Note that in this case, any 693 previously-parsed JSON texts will be lost. 694 695 Example: Parse some JSON arrays/objects in a given string and return 696 them. 697 698 my @objs = JSON::XS->new->incr_parse ("[5][7][1,2]"); 699 700 $lvalue_string = $json->incr_text 701 This method returns the currently stored JSON fragment as an lvalue, 702 that is, you can manipulate it. This *only* works when a preceding 703 call to "incr_parse" in *scalar context* successfully returned an 704 object. Under all other circumstances you must not call this 705 function (I mean it. although in simple tests it might actually 706 work, it *will* fail under real world conditions). As a special 707 exception, you can also call this method before having parsed 708 anything. 709 710 That means you can only use this function to look at or manipulate 711 text before or after complete JSON objects, not while the parser is 712 in the middle of parsing a JSON object. 713 714 This function is useful in two cases: a) finding the trailing text 715 after a JSON object or b) parsing multiple JSON objects separated by 716 non-JSON text (such as commas). 717 718 $json->incr_skip 719 This will reset the state of the incremental parser and will remove 720 the parsed text from the input buffer so far. This is useful after 721 "incr_parse" died, in which case the input buffer and incremental 722 parser state is left unchanged, to skip the text parsed so far and 723 to reset the parse state. 724 725 The difference to "incr_reset" is that only text until the parse 726 error occurred is removed. 727 728 $json->incr_reset 729 This completely resets the incremental parser, that is, after this 730 call, it will be as if the parser had never parsed anything. 731 732 This is useful if you want to repeatedly parse JSON objects and want 733 to ignore any trailing data, which means you have to reset the 734 parser after each successful decode. 735 736 LIMITATIONS 737 The incremental parser is a non-exact parser: it works by gathering as 738 much text as possible that *could* be a valid JSON text, followed by 739 trying to decode it. 740 741 That means it sometimes needs to read more data than strictly necessary 742 to diagnose an invalid JSON text. For example, after parsing the 743 following fragment, the parser *could* stop with an error, as this 744 fragment *cannot* be the beginning of a valid JSON text: 745 746 [, 747 748 In reality, hopwever, the parser might continue to read data until a 749 length limit is exceeded or it finds a closing bracket. 750 751 EXAMPLES 752 Some examples will make all this clearer. First, a simple example that 753 works similarly to "decode_prefix": We want to decode the JSON object at 754 the start of a string and identify the portion after the JSON object: 755 756 my $text = "[1,2,3] hello"; 757 758 my $json = new JSON::XS; 759 760 my $obj = $json->incr_parse ($text) 761 or die "expected JSON object or array at beginning of string"; 762 763 my $tail = $json->incr_text; 764 # $tail now contains " hello" 765 766 Easy, isn't it? 767 768 Now for a more complicated example: Imagine a hypothetical protocol 769 where you read some requests from a TCP stream, and each request is a 770 JSON array, without any separation between them (in fact, it is often 771 useful to use newlines as "separators", as these get interpreted as 772 whitespace at the start of the JSON text, which makes it possible to 773 test said protocol with "telnet"...). 774 775 Here is how you'd do it (it is trivial to write this in an event-based 776 manner): 777 778 my $json = new JSON::XS; 779 780 # read some data from the socket 781 while (sysread $socket, my $buf, 4096) { 782 783 # split and decode as many requests as possible 784 for my $request ($json->incr_parse ($buf)) { 785 # act on the $request 786 } 787 } 788 789 Another complicated example: Assume you have a string with JSON objects 790 or arrays, all separated by (optional) comma characters (e.g. "[1],[2], 791 [3]"). To parse them, we have to skip the commas between the JSON texts, 792 and here is where the lvalue-ness of "incr_text" comes in useful: 793 794 my $text = "[1],[2], [3]"; 795 my $json = new JSON::XS; 796 797 # void context, so no parsing done 798 $json->incr_parse ($text); 799 800 # now extract as many objects as possible. note the 801 # use of scalar context so incr_text can be called. 802 while (my $obj = $json->incr_parse) { 803 # do something with $obj 804 805 # now skip the optional comma 806 $json->incr_text =~ s/^ \s* , //x; 807 } 808 809 Now lets go for a very complex example: Assume that you have a gigantic 810 JSON array-of-objects, many gigabytes in size, and you want to parse it, 811 but you cannot load it into memory fully (this has actually happened in 812 the real world :). 813 814 Well, you lost, you have to implement your own JSON parser. But JSON::XS 815 can still help you: You implement a (very simple) array parser and let 816 JSON decode the array elements, which are all full JSON objects on their 817 own (this wouldn't work if the array elements could be JSON numbers, for 818 example): 819 820 my $json = new JSON::XS; 821 822 # open the monster 823 open my $fh, "<bigfile.json" 824 or die "bigfile: $!"; 825 826 # first parse the initial "[" 827 for (;;) { 828 sysread $fh, my $buf, 65536 829 or die "read error: $!"; 830 $json->incr_parse ($buf); # void context, so no parsing 831 832 # Exit the loop once we found and removed(!) the initial "[". 833 # In essence, we are (ab-)using the $json object as a simple scalar 834 # we append data to. 835 last if $json->incr_text =~ s/^ \s* \[ //x; 836 } 837 838 # now we have the skipped the initial "[", so continue 839 # parsing all the elements. 840 for (;;) { 841 # in this loop we read data until we got a single JSON object 842 for (;;) { 843 if (my $obj = $json->incr_parse) { 844 # do something with $obj 845 last; 846 } 847 848 # add more data 849 sysread $fh, my $buf, 65536 850 or die "read error: $!"; 851 $json->incr_parse ($buf); # void context, so no parsing 852 } 853 854 # in this loop we read data until we either found and parsed the 855 # separating "," between elements, or the final "]" 856 for (;;) { 857 # first skip whitespace 858 $json->incr_text =~ s/^\s*//; 859 860 # if we find "]", we are done 861 if ($json->incr_text =~ s/^\]//) { 862 print "finished.\n"; 863 exit; 864 } 865 866 # if we find ",", we can continue with the next element 867 if ($json->incr_text =~ s/^,//) { 868 last; 869 } 870 871 # if we find anything else, we have a parse error! 872 if (length $json->incr_text) { 873 die "parse error near ", $json->incr_text; 874 } 875 876 # else add more data 877 sysread $fh, my $buf, 65536 878 or die "read error: $!"; 879 $json->incr_parse ($buf); # void context, so no parsing 880 } 881 882 This is a complex example, but most of the complexity comes from the 883 fact that we are trying to be correct (bear with me if I am wrong, I 884 never ran the above example :). 885 886MAPPING 887 This section describes how JSON::XS maps Perl values to JSON values and 888 vice versa. These mappings are designed to "do the right thing" in most 889 circumstances automatically, preserving round-tripping characteristics 890 (what you put in comes out as something equivalent). 891 892 For the more enlightened: note that in the following descriptions, 893 lowercase *perl* refers to the Perl interpreter, while uppercase *Perl* 894 refers to the abstract Perl language itself. 895 896 JSON -> PERL 897 object 898 A JSON object becomes a reference to a hash in Perl. No ordering of 899 object keys is preserved (JSON does not preserve object key ordering 900 itself). 901 902 array 903 A JSON array becomes a reference to an array in Perl. 904 905 string 906 A JSON string becomes a string scalar in Perl - Unicode codepoints 907 in JSON are represented by the same codepoints in the Perl string, 908 so no manual decoding is necessary. 909 910 number 911 A JSON number becomes either an integer, numeric (floating point) or 912 string scalar in perl, depending on its range and any fractional 913 parts. On the Perl level, there is no difference between those as 914 Perl handles all the conversion details, but an integer may take 915 slightly less memory and might represent more values exactly than 916 floating point numbers. 917 918 If the number consists of digits only, JSON::XS will try to 919 represent it as an integer value. If that fails, it will try to 920 represent it as a numeric (floating point) value if that is possible 921 without loss of precision. Otherwise it will preserve the number as 922 a string value (in which case you lose roundtripping ability, as the 923 JSON number will be re-encoded to a JSON string). 924 925 Numbers containing a fractional or exponential part will always be 926 represented as numeric (floating point) values, possibly at a loss 927 of precision (in which case you might lose perfect roundtripping 928 ability, but the JSON number will still be re-encoded as a JSON 929 number). 930 931 Note that precision is not accuracy - binary floating point values 932 cannot represent most decimal fractions exactly, and when converting 933 from and to floating point, JSON::XS only guarantees precision up to 934 but not including the least significant bit. 935 936 true, false 937 These JSON atoms become "Types::Serialiser::true" and 938 "Types::Serialiser::false", respectively. They are overloaded to act 939 almost exactly like the numbers 1 and 0. You can check whether a 940 scalar is a JSON boolean by using the "Types::Serialiser::is_bool" 941 function (after "use Types::Serialier", of course). 942 943 null 944 A JSON null atom becomes "undef" in Perl. 945 946 shell-style comments ("# *text*") 947 As a nonstandard extension to the JSON syntax that is enabled by the 948 "relaxed" setting, shell-style comments are allowed. They can start 949 anywhere outside strings and go till the end of the line. 950 951 tagged values ("(*tag*)*value*"). 952 Another nonstandard extension to the JSON syntax, enabled with the 953 "allow_tags" setting, are tagged values. In this implementation, the 954 *tag* must be a perl package/class name encoded as a JSON string, 955 and the *value* must be a JSON array encoding optional constructor 956 arguments. 957 958 See "OBJECT SERIALISATION", below, for details. 959 960 PERL -> JSON 961 The mapping from Perl to JSON is slightly more difficult, as Perl is a 962 truly typeless language, so we can only guess which JSON type is meant 963 by a Perl value. 964 965 hash references 966 Perl hash references become JSON objects. As there is no inherent 967 ordering in hash keys (or JSON objects), they will usually be 968 encoded in a pseudo-random order. JSON::XS can optionally sort the 969 hash keys (determined by the *canonical* flag), so the same 970 datastructure will serialise to the same JSON text (given same 971 settings and version of JSON::XS), but this incurs a runtime 972 overhead and is only rarely useful, e.g. when you want to compare 973 some JSON text against another for equality. 974 975 array references 976 Perl array references become JSON arrays. 977 978 other references 979 Other unblessed references are generally not allowed and will cause 980 an exception to be thrown, except for references to the integers 0 981 and 1, which get turned into "false" and "true" atoms in JSON. 982 983 Since "JSON::XS" uses the boolean model from Types::Serialiser, you 984 can also "use Types::Serialiser" and then use 985 "Types::Serialiser::false" and "Types::Serialiser::true" to improve 986 readability. 987 988 use Types::Serialiser; 989 encode_json [\0, Types::Serialiser::true] # yields [false,true] 990 991 Types::Serialiser::true, Types::Serialiser::false 992 These special values from the Types::Serialiser module become JSON 993 true and JSON false values, respectively. You can also use "\1" and 994 "\0" directly if you want. 995 996 blessed objects 997 Blessed objects are not directly representable in JSON, but 998 "JSON::XS" allows various ways of handling objects. See "OBJECT 999 SERIALISATION", below, for details. 1000 1001 simple scalars 1002 Simple Perl scalars (any scalar that is not a reference) are the 1003 most difficult objects to encode: JSON::XS will encode undefined 1004 scalars as JSON "null" values, scalars that have last been used in a 1005 string context before encoding as JSON strings, and anything else as 1006 number value: 1007 1008 # dump as number 1009 encode_json [2] # yields [2] 1010 encode_json [-3.0e17] # yields [-3e+17] 1011 my $value = 5; encode_json [$value] # yields [5] 1012 1013 # used as string, so dump as string 1014 print $value; 1015 encode_json [$value] # yields ["5"] 1016 1017 # undef becomes null 1018 encode_json [undef] # yields [null] 1019 1020 You can force the type to be a JSON string by stringifying it: 1021 1022 my $x = 3.1; # some variable containing a number 1023 "$x"; # stringified 1024 $x .= ""; # another, more awkward way to stringify 1025 print $x; # perl does it for you, too, quite often 1026 1027 You can force the type to be a JSON number by numifying it: 1028 1029 my $x = "3"; # some variable containing a string 1030 $x += 0; # numify it, ensuring it will be dumped as a number 1031 $x *= 1; # same thing, the choice is yours. 1032 1033 You can not currently force the type in other, less obscure, ways. 1034 Tell me if you need this capability (but don't forget to explain why 1035 it's needed :). 1036 1037 Note that numerical precision has the same meaning as under Perl (so 1038 binary to decimal conversion follows the same rules as in Perl, 1039 which can differ to other languages). Also, your perl interpreter 1040 might expose extensions to the floating point numbers of your 1041 platform, such as infinities or NaN's - these cannot be represented 1042 in JSON, and it is an error to pass those in. 1043 1044 OBJECT SERIALISATION 1045 As JSON cannot directly represent Perl objects, you have to choose 1046 between a pure JSON representation (without the ability to deserialise 1047 the object automatically again), and a nonstandard extension to the JSON 1048 syntax, tagged values. 1049 1050 SERIALISATION 1051 What happens when "JSON::XS" encounters a Perl object depends on the 1052 "allow_blessed", "convert_blessed" and "allow_tags" settings, which are 1053 used in this order: 1054 1055 1. "allow_tags" is enabled and the object has a "FREEZE" method. 1056 In this case, "JSON::XS" uses the Types::Serialiser object 1057 serialisation protocol to create a tagged JSON value, using a 1058 nonstandard extension to the JSON syntax. 1059 1060 This works by invoking the "FREEZE" method on the object, with the 1061 first argument being the object to serialise, and the second 1062 argument being the constant string "JSON" to distinguish it from 1063 other serialisers. 1064 1065 The "FREEZE" method can return any number of values (i.e. zero or 1066 more). These values and the paclkage/classname of the object will 1067 then be encoded as a tagged JSON value in the following format: 1068 1069 ("classname")[FREEZE return values...] 1070 1071 e.g.: 1072 1073 ("URI")["http://www.google.com/"] 1074 ("MyDate")[2013,10,29] 1075 ("ImageData::JPEG")["Z3...VlCg=="] 1076 1077 For example, the hypothetical "My::Object" "FREEZE" method might use 1078 the objects "type" and "id" members to encode the object: 1079 1080 sub My::Object::FREEZE { 1081 my ($self, $serialiser) = @_; 1082 1083 ($self->{type}, $self->{id}) 1084 } 1085 1086 2. "convert_blessed" is enabled and the object has a "TO_JSON" method. 1087 In this case, the "TO_JSON" method of the object is invoked in 1088 scalar context. It must return a single scalar that can be directly 1089 encoded into JSON. This scalar replaces the object in the JSON text. 1090 1091 For example, the following "TO_JSON" method will convert all URI 1092 objects to JSON strings when serialised. The fatc that these values 1093 originally were URI objects is lost. 1094 1095 sub URI::TO_JSON { 1096 my ($uri) = @_; 1097 $uri->as_string 1098 } 1099 1100 3. "allow_blessed" is enabled. 1101 The object will be serialised as a JSON null value. 1102 1103 4. none of the above 1104 If none of the settings are enabled or the respective methods are 1105 missing, "JSON::XS" throws an exception. 1106 1107 DESERIALISATION 1108 For deserialisation there are only two cases to consider: either 1109 nonstandard tagging was used, in which case "allow_tags" decides, or 1110 objects cannot be automatically be deserialised, in which case you can 1111 use postprocessing or the "filter_json_object" or 1112 "filter_json_single_key_object" callbacks to get some real objects our 1113 of your JSON. 1114 1115 This section only considers the tagged value case: I a tagged JSON 1116 object is encountered during decoding and "allow_tags" is disabled, a 1117 parse error will result (as if tagged values were not part of the 1118 grammar). 1119 1120 If "allow_tags" is enabled, "JSON::XS" will look up the "THAW" method of 1121 the package/classname used during serialisation (it will not attempt to 1122 load the package as a Perl module). If there is no such method, the 1123 decoding will fail with an error. 1124 1125 Otherwise, the "THAW" method is invoked with the classname as first 1126 argument, the constant string "JSON" as second argument, and all the 1127 values from the JSON array (the values originally returned by the 1128 "FREEZE" method) as remaining arguments. 1129 1130 The method must then return the object. While technically you can return 1131 any Perl scalar, you might have to enable the "enable_nonref" setting to 1132 make that work in all cases, so better return an actual blessed 1133 reference. 1134 1135 As an example, let's implement a "THAW" function that regenerates the 1136 "My::Object" from the "FREEZE" example earlier: 1137 1138 sub My::Object::THAW { 1139 my ($class, $serialiser, $type, $id) = @_; 1140 1141 $class->new (type => $type, id => $id) 1142 } 1143 1144ENCODING/CODESET FLAG NOTES 1145 The interested reader might have seen a number of flags that signify 1146 encodings or codesets - "utf8", "latin1" and "ascii". There seems to be 1147 some confusion on what these do, so here is a short comparison: 1148 1149 "utf8" controls whether the JSON text created by "encode" (and expected 1150 by "decode") is UTF-8 encoded or not, while "latin1" and "ascii" only 1151 control whether "encode" escapes character values outside their 1152 respective codeset range. Neither of these flags conflict with each 1153 other, although some combinations make less sense than others. 1154 1155 Care has been taken to make all flags symmetrical with respect to 1156 "encode" and "decode", that is, texts encoded with any combination of 1157 these flag values will be correctly decoded when the same flags are used 1158 - in general, if you use different flag settings while encoding vs. when 1159 decoding you likely have a bug somewhere. 1160 1161 Below comes a verbose discussion of these flags. Note that a "codeset" 1162 is simply an abstract set of character-codepoint pairs, while an 1163 encoding takes those codepoint numbers and *encodes* them, in our case 1164 into octets. Unicode is (among other things) a codeset, UTF-8 is an 1165 encoding, and ISO-8859-1 (= latin 1) and ASCII are both codesets *and* 1166 encodings at the same time, which can be confusing. 1167 1168 "utf8" flag disabled 1169 When "utf8" is disabled (the default), then "encode"/"decode" 1170 generate and expect Unicode strings, that is, characters with high 1171 ordinal Unicode values (> 255) will be encoded as such characters, 1172 and likewise such characters are decoded as-is, no changes to them 1173 will be done, except "(re-)interpreting" them as Unicode codepoints 1174 or Unicode characters, respectively (to Perl, these are the same 1175 thing in strings unless you do funny/weird/dumb stuff). 1176 1177 This is useful when you want to do the encoding yourself (e.g. when 1178 you want to have UTF-16 encoded JSON texts) or when some other layer 1179 does the encoding for you (for example, when printing to a terminal 1180 using a filehandle that transparently encodes to UTF-8 you certainly 1181 do NOT want to UTF-8 encode your data first and have Perl encode it 1182 another time). 1183 1184 "utf8" flag enabled 1185 If the "utf8"-flag is enabled, "encode"/"decode" will encode all 1186 characters using the corresponding UTF-8 multi-byte sequence, and 1187 will expect your input strings to be encoded as UTF-8, that is, no 1188 "character" of the input string must have any value > 255, as UTF-8 1189 does not allow that. 1190 1191 The "utf8" flag therefore switches between two modes: disabled means 1192 you will get a Unicode string in Perl, enabled means you get a UTF-8 1193 encoded octet/binary string in Perl. 1194 1195 "latin1" or "ascii" flags enabled 1196 With "latin1" (or "ascii") enabled, "encode" will escape characters 1197 with ordinal values > 255 (> 127 with "ascii") and encode the 1198 remaining characters as specified by the "utf8" flag. 1199 1200 If "utf8" is disabled, then the result is also correctly encoded in 1201 those character sets (as both are proper subsets of Unicode, meaning 1202 that a Unicode string with all character values < 256 is the same 1203 thing as a ISO-8859-1 string, and a Unicode string with all 1204 character values < 128 is the same thing as an ASCII string in 1205 Perl). 1206 1207 If "utf8" is enabled, you still get a correct UTF-8-encoded string, 1208 regardless of these flags, just some more characters will be escaped 1209 using "\uXXXX" then before. 1210 1211 Note that ISO-8859-1-*encoded* strings are not compatible with UTF-8 1212 encoding, while ASCII-encoded strings are. That is because the 1213 ISO-8859-1 encoding is NOT a subset of UTF-8 (despite the ISO-8859-1 1214 *codeset* being a subset of Unicode), while ASCII is. 1215 1216 Surprisingly, "decode" will ignore these flags and so treat all 1217 input values as governed by the "utf8" flag. If it is disabled, this 1218 allows you to decode ISO-8859-1- and ASCII-encoded strings, as both 1219 strict subsets of Unicode. If it is enabled, you can correctly 1220 decode UTF-8 encoded strings. 1221 1222 So neither "latin1" nor "ascii" are incompatible with the "utf8" 1223 flag - they only govern when the JSON output engine escapes a 1224 character or not. 1225 1226 The main use for "latin1" is to relatively efficiently store binary 1227 data as JSON, at the expense of breaking compatibility with most 1228 JSON decoders. 1229 1230 The main use for "ascii" is to force the output to not contain 1231 characters with values > 127, which means you can interpret the 1232 resulting string as UTF-8, ISO-8859-1, ASCII, KOI8-R or most about 1233 any character set and 8-bit-encoding, and still get the same data 1234 structure back. This is useful when your channel for JSON transfer 1235 is not 8-bit clean or the encoding might be mangled in between (e.g. 1236 in mail), and works because ASCII is a proper subset of most 8-bit 1237 and multibyte encodings in use in the world. 1238 1239 JSON and ECMAscript 1240 JSON syntax is based on how literals are represented in javascript (the 1241 not-standardised predecessor of ECMAscript) which is presumably why it 1242 is called "JavaScript Object Notation". 1243 1244 However, JSON is not a subset (and also not a superset of course) of 1245 ECMAscript (the standard) or javascript (whatever browsers actually 1246 implement). 1247 1248 If you want to use javascript's "eval" function to "parse" JSON, you 1249 might run into parse errors for valid JSON texts, or the resulting data 1250 structure might not be queryable: 1251 1252 One of the problems is that U+2028 and U+2029 are valid characters 1253 inside JSON strings, but are not allowed in ECMAscript string literals, 1254 so the following Perl fragment will not output something that can be 1255 guaranteed to be parsable by javascript's "eval": 1256 1257 use JSON::XS; 1258 1259 print encode_json [chr 0x2028]; 1260 1261 The right fix for this is to use a proper JSON parser in your javascript 1262 programs, and not rely on "eval" (see for example Douglas Crockford's 1263 json2.js parser). 1264 1265 If this is not an option, you can, as a stop-gap measure, simply encode 1266 to ASCII-only JSON: 1267 1268 use JSON::XS; 1269 1270 print JSON::XS->new->ascii->encode ([chr 0x2028]); 1271 1272 Note that this will enlarge the resulting JSON text quite a bit if you 1273 have many non-ASCII characters. You might be tempted to run some regexes 1274 to only escape U+2028 and U+2029, e.g.: 1275 1276 # DO NOT USE THIS! 1277 my $json = JSON::XS->new->utf8->encode ([chr 0x2028]); 1278 $json =~ s/\xe2\x80\xa8/\\u2028/g; # escape U+2028 1279 $json =~ s/\xe2\x80\xa9/\\u2029/g; # escape U+2029 1280 print $json; 1281 1282 Note that *this is a bad idea*: the above only works for U+2028 and 1283 U+2029 and thus only for fully ECMAscript-compliant parsers. Many 1284 existing javascript implementations, however, have issues with other 1285 characters as well - using "eval" naively simply *will* cause problems. 1286 1287 Another problem is that some javascript implementations reserve some 1288 property names for their own purposes (which probably makes them 1289 non-ECMAscript-compliant). For example, Iceweasel reserves the 1290 "__proto__" property name for its own purposes. 1291 1292 If that is a problem, you could parse try to filter the resulting JSON 1293 output for these property strings, e.g.: 1294 1295 $json =~ s/"__proto__"\s*:/"__proto__renamed":/g; 1296 1297 This works because "__proto__" is not valid outside of strings, so every 1298 occurrence of ""__proto__"\s*:" must be a string used as property name. 1299 1300 If you know of other incompatibilities, please let me know. 1301 1302 JSON and YAML 1303 You often hear that JSON is a subset of YAML. This is, however, a mass 1304 hysteria(*) and very far from the truth (as of the time of this 1305 writing), so let me state it clearly: *in general, there is no way to 1306 configure JSON::XS to output a data structure as valid YAML* that works 1307 in all cases. 1308 1309 If you really must use JSON::XS to generate YAML, you should use this 1310 algorithm (subject to change in future versions): 1311 1312 my $to_yaml = JSON::XS->new->utf8->space_after (1); 1313 my $yaml = $to_yaml->encode ($ref) . "\n"; 1314 1315 This will *usually* generate JSON texts that also parse as valid YAML. 1316 Please note that YAML has hardcoded limits on (simple) object key 1317 lengths that JSON doesn't have and also has different and incompatible 1318 unicode character escape syntax, so you should make sure that your hash 1319 keys are noticeably shorter than the 1024 "stream characters" YAML 1320 allows and that you do not have characters with codepoint values outside 1321 the Unicode BMP (basic multilingual page). YAML also does not allow "\/" 1322 sequences in strings (which JSON::XS does not *currently* generate, but 1323 other JSON generators might). 1324 1325 There might be other incompatibilities that I am not aware of (or the 1326 YAML specification has been changed yet again - it does so quite often). 1327 In general you should not try to generate YAML with a JSON generator or 1328 vice versa, or try to parse JSON with a YAML parser or vice versa: 1329 chances are high that you will run into severe interoperability problems 1330 when you least expect it. 1331 1332 (*) I have been pressured multiple times by Brian Ingerson (one of the 1333 authors of the YAML specification) to remove this paragraph, despite 1334 him acknowledging that the actual incompatibilities exist. As I was 1335 personally bitten by this "JSON is YAML" lie, I refused and said I 1336 will continue to educate people about these issues, so others do not 1337 run into the same problem again and again. After this, Brian called 1338 me a (quote)*complete and worthless idiot*(unquote). 1339 1340 In my opinion, instead of pressuring and insulting people who 1341 actually clarify issues with YAML and the wrong statements of some 1342 of its proponents, I would kindly suggest reading the JSON spec 1343 (which is not that difficult or long) and finally make YAML 1344 compatible to it, and educating users about the changes, instead of 1345 spreading lies about the real compatibility for many *years* and 1346 trying to silence people who point out that it isn't true. 1347 1348 Addendum/2009: the YAML 1.2 spec is still incompatible with JSON, 1349 even though the incompatibilities have been documented (and are 1350 known to Brian) for many years and the spec makes explicit claims 1351 that YAML is a superset of JSON. It would be so easy to fix, but 1352 apparently, bullying people and corrupting userdata is so much 1353 easier. 1354 1355 SPEED 1356 It seems that JSON::XS is surprisingly fast, as shown in the following 1357 tables. They have been generated with the help of the "eg/bench" program 1358 in the JSON::XS distribution, to make it easy to compare on your own 1359 system. 1360 1361 First comes a comparison between various modules using a very short 1362 single-line JSON string (also available at 1363 <http://dist.schmorp.de/misc/json/short.json>). 1364 1365 {"method": "handleMessage", "params": ["user1", 1366 "we were just talking"], "id": null, "array":[1,11,234,-5,1e5,1e7, 1367 1, 0]} 1368 1369 It shows the number of encodes/decodes per second (JSON::XS uses the 1370 functional interface, while JSON::XS/2 uses the OO interface with 1371 pretty-printing and hashkey sorting enabled, JSON::XS/3 enables shrink. 1372 JSON::DWIW/DS uses the deserialise function, while JSON::DWIW::FJ uses 1373 the from_json method). Higher is better: 1374 1375 module | encode | decode | 1376 --------------|------------|------------| 1377 JSON::DWIW/DS | 86302.551 | 102300.098 | 1378 JSON::DWIW/FJ | 86302.551 | 75983.768 | 1379 JSON::PP | 15827.562 | 6638.658 | 1380 JSON::Syck | 63358.066 | 47662.545 | 1381 JSON::XS | 511500.488 | 511500.488 | 1382 JSON::XS/2 | 291271.111 | 388361.481 | 1383 JSON::XS/3 | 361577.931 | 361577.931 | 1384 Storable | 66788.280 | 265462.278 | 1385 --------------+------------+------------+ 1386 1387 That is, JSON::XS is almost six times faster than JSON::DWIW on 1388 encoding, about five times faster on decoding, and over thirty to 1389 seventy times faster than JSON's pure perl implementation. It also 1390 compares favourably to Storable for small amounts of data. 1391 1392 Using a longer test string (roughly 18KB, generated from Yahoo! Locals 1393 search API (<http://dist.schmorp.de/misc/json/long.json>). 1394 1395 module | encode | decode | 1396 --------------|------------|------------| 1397 JSON::DWIW/DS | 1647.927 | 2673.916 | 1398 JSON::DWIW/FJ | 1630.249 | 2596.128 | 1399 JSON::PP | 400.640 | 62.311 | 1400 JSON::Syck | 1481.040 | 1524.869 | 1401 JSON::XS | 20661.596 | 9541.183 | 1402 JSON::XS/2 | 10683.403 | 9416.938 | 1403 JSON::XS/3 | 20661.596 | 9400.054 | 1404 Storable | 19765.806 | 10000.725 | 1405 --------------+------------+------------+ 1406 1407 Again, JSON::XS leads by far (except for Storable which non-surprisingly 1408 decodes a bit faster). 1409 1410 On large strings containing lots of high Unicode characters, some 1411 modules (such as JSON::PC) seem to decode faster than JSON::XS, but the 1412 result will be broken due to missing (or wrong) Unicode handling. Others 1413 refuse to decode or encode properly, so it was impossible to prepare a 1414 fair comparison table for that case. 1415 1416SECURITY CONSIDERATIONS 1417 When you are using JSON in a protocol, talking to untrusted potentially 1418 hostile creatures requires relatively few measures. 1419 1420 First of all, your JSON decoder should be secure, that is, should not 1421 have any buffer overflows. Obviously, this module should ensure that and 1422 I am trying hard on making that true, but you never know. 1423 1424 Second, you need to avoid resource-starving attacks. That means you 1425 should limit the size of JSON texts you accept, or make sure then when 1426 your resources run out, that's just fine (e.g. by using a separate 1427 process that can crash safely). The size of a JSON text in octets or 1428 characters is usually a good indication of the size of the resources 1429 required to decode it into a Perl structure. While JSON::XS can check 1430 the size of the JSON text, it might be too late when you already have it 1431 in memory, so you might want to check the size before you accept the 1432 string. 1433 1434 Third, JSON::XS recurses using the C stack when decoding objects and 1435 arrays. The C stack is a limited resource: for instance, on my amd64 1436 machine with 8MB of stack size I can decode around 180k nested arrays 1437 but only 14k nested JSON objects (due to perl itself recursing deeply on 1438 croak to free the temporary). If that is exceeded, the program crashes. 1439 To be conservative, the default nesting limit is set to 512. If your 1440 process has a smaller stack, you should adjust this setting accordingly 1441 with the "max_depth" method. 1442 1443 Something else could bomb you, too, that I forgot to think of. In that 1444 case, you get to keep the pieces. I am always open for hints, though... 1445 1446 Also keep in mind that JSON::XS might leak contents of your Perl data 1447 structures in its error messages, so when you serialise sensitive 1448 information you might want to make sure that exceptions thrown by 1449 JSON::XS will not end up in front of untrusted eyes. 1450 1451 If you are using JSON::XS to return packets to consumption by JavaScript 1452 scripts in a browser you should have a look at 1453 <http://blog.archive.jpsykes.com/47/practical-csrf-and-json-security/> 1454 to see whether you are vulnerable to some common attack vectors (which 1455 really are browser design bugs, but it is still you who will have to 1456 deal with it, as major browser developers care only for features, not 1457 about getting security right). 1458 1459 "OLD" VS. "NEW" JSON (RFC4627 VS. RFC7159) 1460 JSON originally required JSON texts to represent an array or object - 1461 scalar values were explicitly not allowed. This has changed, and 1462 versions of JSON::XS beginning with 4.0 reflect this by allowing scalar 1463 values by default. 1464 1465 One reason why one might not want this is that this removes a 1466 fundamental property of JSON texts, namely that they are self-delimited 1467 and self-contained, or in other words, you could take any number of 1468 "old" JSON texts and paste them together, and the result would be 1469 unambiguously parseable: 1470 1471 [1,3]{"k":5}[][null] # four JSON texts, without doubt 1472 1473 By allowing scalars, this property is lost: in the following example, is 1474 this one JSON text (the number 12) or two JSON texts (the numbers 1 and 1475 2): 1476 1477 12 # could be 12, or 1 and 2 1478 1479 Another lost property of "old" JSON is that no lookahead is required to 1480 know the end of a JSON text, i.e. the JSON text definitely ended at the 1481 last "]" or "}" character, there was no need to read extra characters. 1482 1483 For example, a viable network protocol with "old" JSON was to simply 1484 exchange JSON texts without delimiter. For "new" JSON, you have to use a 1485 suitable delimiter (such as a newline) after every JSON text or ensure 1486 you never encode/decode scalar values. 1487 1488 Most protocols do work by only transferring arrays or objects, and the 1489 easiest way to avoid problems with the "new" JSON definition is to 1490 explicitly disallow scalar values in your encoder and decoder: 1491 1492 $json_coder = JSON::XS->new->allow_nonref (0) 1493 1494 This is a somewhat unhappy situation, and the blame can fully be put on 1495 JSON's inmventor, Douglas Crockford, who unilaterally changed the format 1496 in 2006 without consulting the IETF, forcing the IETF to either fork the 1497 format or go with it (as I was told, the IETF wasn't amused). 1498 1499RELATIONSHIP WITH I-JSON 1500 JSON is a somewhat sloppily-defined format - it carries around obvious 1501 Javascript baggage, such as not really defining number range, probably 1502 because Javascript only has one type of numbers: IEEE 64 bit floats 1503 ("binary64"). 1504 1505 For this reaosn, RFC7493 defines "Internet JSON", which is a restricted 1506 subset of JSON that is supposedly more interoperable on the internet. 1507 1508 While "JSON::XS" does not offer specific support for I-JSON, it of 1509 course accepts valid I-JSON and by default implements some of the 1510 limitations of I-JSON, such as parsing numbers as perl numbers, which 1511 are usually a superset of binary64 numbers. 1512 1513 To generate I-JSON, follow these rules: 1514 1515 * always generate UTF-8 1516 1517 I-JSON must be encoded in UTF-8, the default for "encode_json". 1518 1519 * numbers should be within IEEE 754 binary64 range 1520 1521 Basically all existing perl installations use binary64 to represent 1522 floating point numbers, so all you need to do is to avoid large 1523 integers. 1524 1525 * objects must not have duplicate keys 1526 1527 This is trivially done, as "JSON::XS" does not allow duplicate keys. 1528 1529 * do not generate scalar JSON texts, use "->allow_nonref (0)" 1530 1531 I-JSON strongly requests you to only encode arrays and objects into 1532 JSON. 1533 1534 * times should be strings in ISO 8601 format 1535 1536 There are a myriad of modules on CPAN dealing with ISO 8601 - search 1537 for "ISO8601" on CPAN and use one. 1538 1539 * encode binary data as base64 1540 1541 While it's tempting to just dump binary data as a string (and let 1542 "JSON::XS" do the escaping), for I-JSON, it's *recommended* to 1543 encode binary data as base64. 1544 1545 There are some other considerations - read RFC7493 for the details if 1546 interested. 1547 1548INTEROPERABILITY WITH OTHER MODULES 1549 "JSON::XS" uses the Types::Serialiser module to provide boolean 1550 constants. That means that the JSON true and false values will be 1551 comaptible to true and false values of other modules that do the same, 1552 such as JSON::PP and CBOR::XS. 1553 1554INTEROPERABILITY WITH OTHER JSON DECODERS 1555 As long as you only serialise data that can be directly expressed in 1556 JSON, "JSON::XS" is incapable of generating invalid JSON output (modulo 1557 bugs, but "JSON::XS" has found more bugs in the official JSON testsuite 1558 (1) than the official JSON testsuite has found in "JSON::XS" (0)). 1559 1560 When you have trouble decoding JSON generated by this module using other 1561 decoders, then it is very likely that you have an encoding mismatch or 1562 the other decoder is broken. 1563 1564 When decoding, "JSON::XS" is strict by default and will likely catch all 1565 errors. There are currently two settings that change this: "relaxed" 1566 makes "JSON::XS" accept (but not generate) some non-standard extensions, 1567 and "allow_tags" will allow you to encode and decode Perl objects, at 1568 the cost of not outputting valid JSON anymore. 1569 1570 TAGGED VALUE SYNTAX AND STANDARD JSON EN/DECODERS 1571 When you use "allow_tags" to use the extended (and also nonstandard and 1572 invalid) JSON syntax for serialised objects, and you still want to 1573 decode the generated When you want to serialise objects, you can run a 1574 regex to replace the tagged syntax by standard JSON arrays (it only 1575 works for "normal" package names without comma, newlines or single 1576 colons). First, the readable Perl version: 1577 1578 # if your FREEZE methods return no values, you need this replace first: 1579 $json =~ s/\( \s* (" (?: [^\\":,]+|\\.|::)* ") \s* \) \s* \[\s*\]/[$1]/gx; 1580 1581 # this works for non-empty constructor arg lists: 1582 $json =~ s/\( \s* (" (?: [^\\":,]+|\\.|::)* ") \s* \) \s* \[/[$1,/gx; 1583 1584 And here is a less readable version that is easy to adapt to other 1585 languages: 1586 1587 $json =~ s/\(\s*("([^\\":,]+|\\.|::)*")\s*\)\s*\[/[$1,/g; 1588 1589 Here is an ECMAScript version (same regex): 1590 1591 json = json.replace (/\(\s*("([^\\":,]+|\\.|::)*")\s*\)\s*\[/g, "[$1,"); 1592 1593 Since this syntax converts to standard JSON arrays, it might be hard to 1594 distinguish serialised objects from normal arrays. You can prepend a 1595 "magic number" as first array element to reduce chances of a collision: 1596 1597 $json =~ s/\(\s*("([^\\":,]+|\\.|::)*")\s*\)\s*\[/["XU1peReLzT4ggEllLanBYq4G9VzliwKF",$1,/g; 1598 1599 And after decoding the JSON text, you could walk the data structure 1600 looking for arrays with a first element of 1601 "XU1peReLzT4ggEllLanBYq4G9VzliwKF". 1602 1603 The same approach can be used to create the tagged format with another 1604 encoder. First, you create an array with the magic string as first 1605 member, the classname as second, and constructor arguments last, encode 1606 it as part of your JSON structure, and then: 1607 1608 $json =~ s/\[\s*"XU1peReLzT4ggEllLanBYq4G9VzliwKF"\s*,\s*("([^\\":,]+|\\.|::)*")\s*,/($1)[/g; 1609 1610 Again, this has some limitations - the magic string must not be encoded 1611 with character escapes, and the constructor arguments must be non-empty. 1612 1613(I-)THREADS 1614 This module is *not* guaranteed to be ithread (or MULTIPLICITY-) safe 1615 and there are no plans to change this. Note that perl's builtin 1616 so-called threads/ithreads are officially deprecated and should not be 1617 used. 1618 1619THE PERILS OF SETLOCALE 1620 Sometimes people avoid the Perl locale support and directly call the 1621 system's setlocale function with "LC_ALL". 1622 1623 This breaks both perl and modules such as JSON::XS, as stringification 1624 of numbers no longer works correctly (e.g. "$x = 0.1; print "$x"+1" 1625 might print 1, and JSON::XS might output illegal JSON as JSON::XS relies 1626 on perl to stringify numbers). 1627 1628 The solution is simple: don't call "setlocale", or use it for only those 1629 categories you need, such as "LC_MESSAGES" or "LC_CTYPE". 1630 1631 If you need "LC_NUMERIC", you should enable it only around the code that 1632 actually needs it (avoiding stringification of numbers), and restore it 1633 afterwards. 1634 1635SOME HISTORY 1636 At the time this module was created there already were a number of JSON 1637 modules available on CPAN, so what was the reason to write yet another 1638 JSON module? While it seems there are many JSON modules, none of them 1639 correctly handled all corner cases, and in most cases their maintainers 1640 are unresponsive, gone missing, or not listening to bug reports for 1641 other reasons. 1642 1643 Beginning with version 2.0 of the JSON module, when both JSON and 1644 JSON::XS are installed, then JSON will fall back on JSON::XS (this can 1645 be overridden) with no overhead due to emulation (by inheriting 1646 constructor and methods). If JSON::XS is not available, it will fall 1647 back to the compatible JSON::PP module as backend, so using JSON instead 1648 of JSON::XS gives you a portable JSON API that can be fast when you need 1649 it and doesn't require a C compiler when that is a problem. 1650 1651 Somewhere around version 3, this module was forked into 1652 "Cpanel::JSON::XS", because its maintainer had serious trouble 1653 understanding JSON and insisted on a fork with many bugs "fixed" that 1654 weren't actually bugs, while spreading FUD about this module without 1655 actually giving any details on his accusations. You be the judge, but in 1656 my personal opinion, if you want quality, you will stay away from 1657 dangerous forks like that. 1658 1659BUGS 1660 While the goal of this module is to be correct, that unfortunately does 1661 not mean it's bug-free, only that I think its design is bug-free. If you 1662 keep reporting bugs they will be fixed swiftly, though. 1663 1664 Please refrain from using rt.cpan.org or any other bug reporting 1665 service. I put the contact address into my modules for a reason. 1666 1667SEE ALSO 1668 The json_xs command line utility for quick experiments. 1669 1670AUTHOR 1671 Marc Lehmann <schmorp@schmorp.de> 1672 http://home.schmorp.de/ 1673 1674