… | |
… | |
9 | # exported functions, they croak on error |
9 | # exported functions, they croak on error |
10 | # and expect/generate UTF-8 |
10 | # and expect/generate UTF-8 |
11 | |
11 | |
12 | $utf8_encoded_json_text = to_json $perl_hash_or_arrayref; |
12 | $utf8_encoded_json_text = to_json $perl_hash_or_arrayref; |
13 | $perl_hash_or_arrayref = from_json $utf8_encoded_json_text; |
13 | $perl_hash_or_arrayref = from_json $utf8_encoded_json_text; |
14 | |
|
|
15 | # objToJson and jsonToObj aliases to to_json and from_json |
|
|
16 | # are exported for compatibility to the JSON module, |
|
|
17 | # but should not be used in new code. |
|
|
18 | |
14 | |
19 | # OO-interface |
15 | # OO-interface |
20 | |
16 | |
21 | $coder = JSON::XS->new->ascii->pretty->allow_nonref; |
17 | $coder = JSON::XS->new->ascii->pretty->allow_nonref; |
22 | $pretty_printed_unencoded = $coder->encode ($perl_scalar); |
18 | $pretty_printed_unencoded = $coder->encode ($perl_scalar); |
… | |
… | |
85 | |
81 | |
86 | package JSON::XS; |
82 | package JSON::XS; |
87 | |
83 | |
88 | use strict; |
84 | use strict; |
89 | |
85 | |
90 | our $VERSION = '1.4'; |
86 | our $VERSION = '1.43'; |
91 | our @ISA = qw(Exporter); |
87 | our @ISA = qw(Exporter); |
92 | |
88 | |
93 | our @EXPORT = qw(to_json from_json objToJson jsonToObj); |
89 | our @EXPORT = qw(to_json from_json); |
94 | |
90 | |
95 | use Exporter; |
91 | use Exporter; |
96 | use XSLoader; |
92 | use XSLoader; |
97 | |
93 | |
98 | =head1 FUNCTIONAL INTERFACE |
94 | =head1 FUNCTIONAL INTERFACE |
… | |
… | |
341 | |
337 | |
342 | The C<TO_JSON> method may safely call die if it wants. If C<TO_JSON> |
338 | The C<TO_JSON> method may safely call die if it wants. If C<TO_JSON> |
343 | returns other blessed objects, those will be handled in the same |
339 | returns other blessed objects, those will be handled in the same |
344 | way. C<TO_JSON> must take care of not causing an endless recursion cycle |
340 | way. C<TO_JSON> must take care of not causing an endless recursion cycle |
345 | (== crash) in this case. The name of C<TO_JSON> was chosen because other |
341 | (== crash) in this case. The name of C<TO_JSON> was chosen because other |
346 | methods called by the Perl core (== not the user of the object) are |
342 | methods called by the Perl core (== not by the user of the object) are |
347 | usually in upper case letters and to avoid collisions with the C<to_json> |
343 | usually in upper case letters and to avoid collisions with the C<to_json> |
348 | function. |
344 | function. |
349 | |
345 | |
|
|
346 | This setting does not yet influence C<decode> in any way, but in the |
|
|
347 | future, global hooks might get installed that influence C<decode> and are |
|
|
348 | enabled by this setting. |
|
|
349 | |
350 | If C<$enable> is false, then the C<allow_blessed> setting will decide what |
350 | If C<$enable> is false, then the C<allow_blessed> setting will decide what |
351 | to do when a blessed object is found. |
351 | to do when a blessed object is found. |
|
|
352 | |
|
|
353 | =item $json = $json->filter_json_object ([$coderef->($hashref)]) |
|
|
354 | |
|
|
355 | When C<$coderef> is specified, it will be called from C<decode> each |
|
|
356 | time it decodes a JSON object. The only argument is a reference to the |
|
|
357 | newly-created hash. If the code references returns a single scalar (which |
|
|
358 | need not be a reference), this value (i.e. a copy of that scalar to avoid |
|
|
359 | aliasing) is inserted into the deserialised data structure. If it returns |
|
|
360 | an empty list (NOTE: I<not> C<undef>, which is a valid scalar), the |
|
|
361 | original deserialised hash will be inserted. This setting can slow down |
|
|
362 | decoding considerably. |
|
|
363 | |
|
|
364 | When C<$coderef> is omitted or undefined, any existing callback will |
|
|
365 | be removed and C<decode> will not change the deserialised hash in any |
|
|
366 | way. |
|
|
367 | |
|
|
368 | Example, convert all JSON objects into the integer 5: |
|
|
369 | |
|
|
370 | my $js = JSON::XS->new->filter_json_object (sub { 5 }); |
|
|
371 | # returns [5] |
|
|
372 | $js->decode ('[{}]') |
|
|
373 | # throw an exception because allow_nonref is not enabled |
|
|
374 | # so a lone 5 is not allowed. |
|
|
375 | $js->decode ('{"a":1, "b":2}'); |
|
|
376 | |
|
|
377 | =item $json = $json->filter_json_single_key_object ($key [=> $coderef->($value)]) |
|
|
378 | |
|
|
379 | Works remotely similar to C<filter_json_object>, but is only called for |
|
|
380 | JSON objects having a single key named C<$key>. |
|
|
381 | |
|
|
382 | This C<$coderef> is called before the one specified via |
|
|
383 | C<filter_json_object>, if any. It gets passed the single value in the JSON |
|
|
384 | object. If it returns a single value, it will be inserted into the data |
|
|
385 | structure. If it returns nothing (not even C<undef> but the empty list), |
|
|
386 | the callback from C<filter_json_object> will be called next, as if no |
|
|
387 | single-key callback were specified. |
|
|
388 | |
|
|
389 | If C<$coderef> is omitted or undefined, the corresponding callback will be |
|
|
390 | disabled. There can only ever be one callback for a given key. |
|
|
391 | |
|
|
392 | As this callback gets called less often then the C<filter_json_object> |
|
|
393 | one, decoding speed will not usually suffer as much. Therefore, single-key |
|
|
394 | objects make excellent targets to serialise Perl objects into, especially |
|
|
395 | as single-key JSON objects are as close to the type-tagged value concept |
|
|
396 | as JSON gets (its basically an ID/VALUE tuple). Of course, JSON does not |
|
|
397 | support this in any way, so you need to make sure your data never looks |
|
|
398 | like a serialised Perl hash. |
|
|
399 | |
|
|
400 | Typical names for the single object key are C<__class_whatever__>, or |
|
|
401 | C<$__dollars_are_rarely_used__$> or C<}ugly_brace_placement>, or even |
|
|
402 | things like C<__class_md5sum(classname)__>, to reduce the risk of clashing |
|
|
403 | with real hashes. |
|
|
404 | |
|
|
405 | Example, decode JSON objects of the form C<< { "__widget__" => <id> } >> |
|
|
406 | into the corresponding C<< $WIDGET{<id>} >> object: |
|
|
407 | |
|
|
408 | # return whatever is in $WIDGET{5}: |
|
|
409 | JSON::XS |
|
|
410 | ->new |
|
|
411 | ->filter_json_single_key_object (__widget__ => sub { |
|
|
412 | $WIDGET{ $_[0] } |
|
|
413 | }) |
|
|
414 | ->decode ('{"__widget__": 5') |
|
|
415 | |
|
|
416 | # this can be used with a TO_JSON method in some "widget" class |
|
|
417 | # for serialisation to json: |
|
|
418 | sub WidgetBase::TO_JSON { |
|
|
419 | my ($self) = @_; |
|
|
420 | |
|
|
421 | unless ($self->{id}) { |
|
|
422 | $self->{id} = ..get..some..id..; |
|
|
423 | $WIDGET{$self->{id}} = $self; |
|
|
424 | } |
|
|
425 | |
|
|
426 | { __widget__ => $self->{id} } |
|
|
427 | } |
352 | |
428 | |
353 | =item $json = $json->shrink ([$enable]) |
429 | =item $json = $json->shrink ([$enable]) |
354 | |
430 | |
355 | Perl usually over-allocates memory a bit when allocating space for |
431 | Perl usually over-allocates memory a bit when allocating space for |
356 | strings. This flag optionally resizes strings generated by either |
432 | strings. This flag optionally resizes strings generated by either |
… | |
… | |
389 | given character in a string. |
465 | given character in a string. |
390 | |
466 | |
391 | Setting the maximum depth to one disallows any nesting, so that ensures |
467 | Setting the maximum depth to one disallows any nesting, so that ensures |
392 | that the object is only a single hash/object or array. |
468 | that the object is only a single hash/object or array. |
393 | |
469 | |
394 | The argument to C<max_depth> will be rounded up to the next nearest power |
470 | The argument to C<max_depth> will be rounded up to the next highest power |
395 | of two. |
471 | of two. If no argument is given, the highest possible setting will be |
|
|
472 | used, which is rarely useful. |
|
|
473 | |
|
|
474 | See SECURITY CONSIDERATIONS, below, for more info on why this is useful. |
|
|
475 | |
|
|
476 | =item $json = $json->max_size ([$maximum_string_size]) |
|
|
477 | |
|
|
478 | Set the maximum length a JSON text may have (in bytes) where decoding is |
|
|
479 | being attempted. The default is C<0>, meaning no limit. When C<decode> |
|
|
480 | is called on a string longer then this number of characters it will not |
|
|
481 | attempt to decode the string but throw an exception. This setting has no |
|
|
482 | effect on C<encode> (yet). |
|
|
483 | |
|
|
484 | The argument to C<max_size> will be rounded up to the next B<highest> |
|
|
485 | power of two (so may be more than requested). If no argument is given, the |
|
|
486 | limit check will be deactivated (same as when C<0> is specified). |
396 | |
487 | |
397 | See SECURITY CONSIDERATIONS, below, for more info on why this is useful. |
488 | See SECURITY CONSIDERATIONS, below, for more info on why this is useful. |
398 | |
489 | |
399 | =item $json_text = $json->encode ($perl_scalar) |
490 | =item $json_text = $json->encode ($perl_scalar) |
400 | |
491 | |
… | |
… | |
462 | are represented by the same codepoints in the Perl string, so no manual |
553 | are represented by the same codepoints in the Perl string, so no manual |
463 | decoding is necessary. |
554 | decoding is necessary. |
464 | |
555 | |
465 | =item number |
556 | =item number |
466 | |
557 | |
467 | A JSON number becomes either an integer or numeric (floating point) |
558 | A JSON number becomes either an integer, numeric (floating point) or |
468 | scalar in perl, depending on its range and any fractional parts. On the |
559 | string scalar in perl, depending on its range and any fractional parts. On |
469 | Perl level, there is no difference between those as Perl handles all the |
560 | the Perl level, there is no difference between those as Perl handles all |
470 | conversion details, but an integer may take slightly less memory and might |
561 | the conversion details, but an integer may take slightly less memory and |
471 | represent more values exactly than (floating point) numbers. |
562 | might represent more values exactly than (floating point) numbers. |
|
|
563 | |
|
|
564 | If the number consists of digits only, JSON::XS will try to represent |
|
|
565 | it as an integer value. If that fails, it will try to represent it as |
|
|
566 | a numeric (floating point) value if that is possible without loss of |
|
|
567 | precision. Otherwise it will preserve the number as a string value. |
|
|
568 | |
|
|
569 | Numbers containing a fractional or exponential part will always be |
|
|
570 | represented as numeric (floating point) values, possibly at a loss of |
|
|
571 | precision. |
|
|
572 | |
|
|
573 | This might create round-tripping problems as numbers might become strings, |
|
|
574 | but as Perl is typeless there is no other way to do it. |
472 | |
575 | |
473 | =item true, false |
576 | =item true, false |
474 | |
577 | |
475 | These JSON atoms become C<JSON::XS::true> and C<JSON::XS::false>, |
578 | These JSON atoms become C<JSON::XS::true> and C<JSON::XS::false>, |
476 | respectively. They are overloaded to act almost exactly like the numbers |
579 | respectively. They are overloaded to act almost exactly like the numbers |
… | |
… | |
692 | It shows the number of encodes/decodes per second (JSON::XS uses |
795 | It shows the number of encodes/decodes per second (JSON::XS uses |
693 | the functional interface, while JSON::XS/2 uses the OO interface |
796 | the functional interface, while JSON::XS/2 uses the OO interface |
694 | with pretty-printing and hashkey sorting enabled, JSON::XS/3 enables |
797 | with pretty-printing and hashkey sorting enabled, JSON::XS/3 enables |
695 | shrink). Higher is better: |
798 | shrink). Higher is better: |
696 | |
799 | |
|
|
800 | Storable | 15779.925 | 14169.946 | |
|
|
801 | -----------+------------+------------+ |
697 | module | encode | decode | |
802 | module | encode | decode | |
698 | -----------|------------|------------| |
803 | -----------|------------|------------| |
699 | JSON | 7645.468 | 4208.613 | |
804 | JSON | 4990.842 | 4088.813 | |
700 | JSON::DWIW | 40721.398 | 77101.176 | |
805 | JSON::DWIW | 51653.990 | 71575.154 | |
701 | JSON::PC | 65948.176 | 78251.940 | |
806 | JSON::PC | 65948.176 | 74631.744 | |
702 | JSON::Syck | 22844.793 | 26479.192 | |
807 | JSON::PP | 8931.652 | 3817.168 | |
|
|
808 | JSON::Syck | 24877.248 | 27776.848 | |
703 | JSON::XS | 388361.481 | 199728.762 | |
809 | JSON::XS | 388361.481 | 227951.304 | |
704 | JSON::XS/2 | 218453.333 | 192399.266 | |
810 | JSON::XS/2 | 227951.304 | 218453.333 | |
705 | JSON::XS/3 | 338250.323 | 192399.266 | |
811 | JSON::XS/3 | 338250.323 | 218453.333 | |
706 | Storable | 15779.925 | 14169.946 | |
812 | Storable | 16500.016 | 135300.129 | |
707 | -----------+------------+------------+ |
813 | -----------+------------+------------+ |
708 | |
814 | |
709 | That is, JSON::XS is about five times faster than JSON::DWIW on encoding, |
815 | That is, JSON::XS is about five times faster than JSON::DWIW on encoding, |
710 | about three times faster on decoding, and over fourty times faster |
816 | about three times faster on decoding, and over fourty times faster |
711 | than JSON, even with pretty-printing and key sorting. It also compares |
817 | than JSON, even with pretty-printing and key sorting. It also compares |
… | |
… | |
714 | Using a longer test string (roughly 18KB, generated from Yahoo! Locals |
820 | Using a longer test string (roughly 18KB, generated from Yahoo! Locals |
715 | search API (http://nanoref.com/yahooapis/mgPdGg): |
821 | search API (http://nanoref.com/yahooapis/mgPdGg): |
716 | |
822 | |
717 | module | encode | decode | |
823 | module | encode | decode | |
718 | -----------|------------|------------| |
824 | -----------|------------|------------| |
719 | JSON | 254.685 | 37.665 | |
825 | JSON | 55.260 | 34.971 | |
720 | JSON::DWIW | 843.343 | 1049.731 | |
826 | JSON::DWIW | 825.228 | 1082.513 | |
721 | JSON::PC | 3602.116 | 2307.352 | |
827 | JSON::PC | 3571.444 | 2394.829 | |
|
|
828 | JSON::PP | 210.987 | 32.574 | |
722 | JSON::Syck | 505.107 | 787.899 | |
829 | JSON::Syck | 552.551 | 787.544 | |
723 | JSON::XS | 5747.196 | 3690.220 | |
830 | JSON::XS | 5780.463 | 4854.519 | |
724 | JSON::XS/2 | 3968.121 | 3676.634 | |
831 | JSON::XS/2 | 3869.998 | 4798.975 | |
725 | JSON::XS/3 | 6105.246 | 3662.508 | |
832 | JSON::XS/3 | 5862.880 | 4798.975 | |
726 | Storable | 4417.337 | 5285.161 | |
833 | Storable | 4445.002 | 5235.027 | |
727 | -----------+------------+------------+ |
834 | -----------+------------+------------+ |
728 | |
835 | |
729 | Again, JSON::XS leads by far (except for Storable which non-surprisingly |
836 | Again, JSON::XS leads by far (except for Storable which non-surprisingly |
730 | decodes faster). |
837 | decodes faster). |
731 | |
838 | |
… | |
… | |
748 | Second, you need to avoid resource-starving attacks. That means you should |
855 | Second, you need to avoid resource-starving attacks. That means you should |
749 | limit the size of JSON texts you accept, or make sure then when your |
856 | limit the size of JSON texts you accept, or make sure then when your |
750 | resources run out, thats just fine (e.g. by using a separate process that |
857 | resources run out, thats just fine (e.g. by using a separate process that |
751 | can crash safely). The size of a JSON text in octets or characters is |
858 | can crash safely). The size of a JSON text in octets or characters is |
752 | usually a good indication of the size of the resources required to decode |
859 | usually a good indication of the size of the resources required to decode |
753 | it into a Perl structure. |
860 | it into a Perl structure. While JSON::XS can check the size of the JSON |
|
|
861 | text, it might be too late when you already have it in memory, so you |
|
|
862 | might want to check the size before you accept the string. |
754 | |
863 | |
755 | Third, JSON::XS recurses using the C stack when decoding objects and |
864 | Third, JSON::XS recurses using the C stack when decoding objects and |
756 | arrays. The C stack is a limited resource: for instance, on my amd64 |
865 | arrays. The C stack is a limited resource: for instance, on my amd64 |
757 | machine with 8MB of stack size I can decode around 180k nested arrays but |
866 | machine with 8MB of stack size I can decode around 180k nested arrays but |
758 | only 14k nested JSON objects (due to perl itself recursing deeply on croak |
867 | only 14k nested JSON objects (due to perl itself recursing deeply on croak |