… | |
… | |
9 | |
9 | |
10 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
10 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
11 | # define snprintf _snprintf // C compilers have this in stdio.h |
11 | # define snprintf _snprintf // C compilers have this in stdio.h |
12 | #endif |
12 | #endif |
13 | |
13 | |
|
|
14 | // some old perls do not have this, try to make it work, no |
|
|
15 | // guarentees, though. if it breaks, you get to keep the pieces. |
|
|
16 | #ifndef UTF8_MAXBYTES |
|
|
17 | # define UTF8_MAXBYTES 13 |
|
|
18 | #endif |
|
|
19 | |
14 | #define F_ASCII 0x00000001UL |
20 | #define F_ASCII 0x00000001UL |
15 | #define F_LATIN1 0x00000002UL |
21 | #define F_LATIN1 0x00000002UL |
16 | #define F_UTF8 0x00000004UL |
22 | #define F_UTF8 0x00000004UL |
17 | #define F_INDENT 0x00000008UL |
23 | #define F_INDENT 0x00000008UL |
18 | #define F_CANONICAL 0x00000010UL |
24 | #define F_CANONICAL 0x00000010UL |
19 | #define F_SPACE_BEFORE 0x00000020UL |
25 | #define F_SPACE_BEFORE 0x00000020UL |
20 | #define F_SPACE_AFTER 0x00000040UL |
26 | #define F_SPACE_AFTER 0x00000040UL |
21 | #define F_ALLOW_NONREF 0x00000100UL |
27 | #define F_ALLOW_NONREF 0x00000100UL |
22 | #define F_SHRINK 0x00000200UL |
28 | #define F_SHRINK 0x00000200UL |
|
|
29 | #define F_ALLOW_BLESSED 0x00000400UL |
|
|
30 | #define F_CONV_BLESSED 0x00000800UL |
23 | #define F_MAXDEPTH 0xf8000000UL |
31 | #define F_MAXDEPTH 0xf8000000UL |
24 | #define S_MAXDEPTH 27 |
32 | #define S_MAXDEPTH 27 |
|
|
33 | #define F_MAXSIZE 0x01f00000UL |
|
|
34 | #define S_MAXSIZE 20 |
|
|
35 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
25 | |
36 | |
26 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
37 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
27 | |
38 | #define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE )) |
28 | // F_SELFCONVERT? <=> to_json/toJson |
|
|
29 | // F_BLESSED? <=> { $__class__$ => } |
|
|
30 | |
39 | |
31 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
40 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
32 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
41 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
33 | |
42 | |
34 | #define INIT_SIZE 32 // initial scalar size to be allocated |
43 | #define INIT_SIZE 32 // initial scalar size to be allocated |
… | |
… | |
48 | #endif |
57 | #endif |
49 | |
58 | |
50 | #define expect_false(expr) expect ((expr) != 0, 0) |
59 | #define expect_false(expr) expect ((expr) != 0, 0) |
51 | #define expect_true(expr) expect ((expr) != 0, 1) |
60 | #define expect_true(expr) expect ((expr) != 0, 1) |
52 | |
61 | |
53 | static HV *json_stash; // JSON::XS:: |
62 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
|
|
63 | static SV *json_true, *json_false; |
|
|
64 | |
|
|
65 | typedef struct { |
|
|
66 | U32 flags; |
|
|
67 | SV *cb_object, *cb_sk_object; |
|
|
68 | } JSON; |
54 | |
69 | |
55 | ///////////////////////////////////////////////////////////////////////////// |
70 | ///////////////////////////////////////////////////////////////////////////// |
56 | // utility functions |
71 | // utility functions |
57 | |
72 | |
58 | static UV * |
73 | inline void |
59 | SvJSON (SV *sv) |
|
|
60 | { |
|
|
61 | if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) |
|
|
62 | croak ("object is not of type JSON::XS"); |
|
|
63 | |
|
|
64 | return &SvUVX (SvRV (sv)); |
|
|
65 | } |
|
|
66 | |
|
|
67 | static void |
|
|
68 | shrink (SV *sv) |
74 | shrink (SV *sv) |
69 | { |
75 | { |
70 | sv_utf8_downgrade (sv, 1); |
76 | sv_utf8_downgrade (sv, 1); |
71 | if (SvLEN (sv) > SvCUR (sv) + 1) |
77 | if (SvLEN (sv) > SvCUR (sv) + 1) |
72 | { |
78 | { |
… | |
… | |
107 | typedef struct |
113 | typedef struct |
108 | { |
114 | { |
109 | char *cur; // SvPVX (sv) + current output position |
115 | char *cur; // SvPVX (sv) + current output position |
110 | char *end; // SvEND (sv) |
116 | char *end; // SvEND (sv) |
111 | SV *sv; // result scalar |
117 | SV *sv; // result scalar |
112 | U32 flags; // F_* |
118 | JSON json; |
113 | U32 indent; // indentation level |
119 | U32 indent; // indentation level |
114 | U32 maxdepth; // max. indentation/recursion level |
120 | U32 maxdepth; // max. indentation/recursion level |
115 | } enc_t; |
121 | } enc_t; |
116 | |
122 | |
117 | inline void |
123 | inline void |
… | |
… | |
178 | STRLEN clen; |
184 | STRLEN clen; |
179 | UV uch; |
185 | UV uch; |
180 | |
186 | |
181 | if (is_utf8) |
187 | if (is_utf8) |
182 | { |
188 | { |
183 | //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
|
|
184 | uch = decode_utf8 (str, end - str, &clen); |
189 | uch = decode_utf8 (str, end - str, &clen); |
185 | if (clen == (STRLEN)-1) |
190 | if (clen == (STRLEN)-1) |
186 | croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
191 | croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
187 | } |
192 | } |
188 | else |
193 | else |
… | |
… | |
192 | } |
197 | } |
193 | |
198 | |
194 | if (uch > 0x10FFFFUL) |
199 | if (uch > 0x10FFFFUL) |
195 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
200 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
196 | |
201 | |
197 | if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF)) |
202 | if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF)) |
198 | { |
203 | { |
199 | if (uch > 0xFFFFUL) |
204 | if (uch > 0xFFFFUL) |
200 | { |
205 | { |
201 | need (enc, len += 11); |
206 | need (enc, len += 11); |
202 | sprintf (enc->cur, "\\u%04x\\u%04x", |
207 | sprintf (enc->cur, "\\u%04x\\u%04x", |
… | |
… | |
216 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
221 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
217 | } |
222 | } |
218 | |
223 | |
219 | str += clen; |
224 | str += clen; |
220 | } |
225 | } |
221 | else if (enc->flags & F_LATIN1) |
226 | else if (enc->json.flags & F_LATIN1) |
222 | { |
227 | { |
223 | *enc->cur++ = uch; |
228 | *enc->cur++ = uch; |
224 | str += clen; |
229 | str += clen; |
225 | } |
230 | } |
226 | else if (is_utf8) |
231 | else if (is_utf8) |
… | |
… | |
247 | } |
252 | } |
248 | |
253 | |
249 | inline void |
254 | inline void |
250 | encode_indent (enc_t *enc) |
255 | encode_indent (enc_t *enc) |
251 | { |
256 | { |
252 | if (enc->flags & F_INDENT) |
257 | if (enc->json.flags & F_INDENT) |
253 | { |
258 | { |
254 | int spaces = enc->indent * INDENT_STEP; |
259 | int spaces = enc->indent * INDENT_STEP; |
255 | |
260 | |
256 | need (enc, spaces); |
261 | need (enc, spaces); |
257 | memset (enc->cur, ' ', spaces); |
262 | memset (enc->cur, ' ', spaces); |
… | |
… | |
267 | } |
272 | } |
268 | |
273 | |
269 | inline void |
274 | inline void |
270 | encode_nl (enc_t *enc) |
275 | encode_nl (enc_t *enc) |
271 | { |
276 | { |
272 | if (enc->flags & F_INDENT) |
277 | if (enc->json.flags & F_INDENT) |
273 | { |
278 | { |
274 | need (enc, 1); |
279 | need (enc, 1); |
275 | encode_ch (enc, '\n'); |
280 | encode_ch (enc, '\n'); |
276 | } |
281 | } |
277 | } |
282 | } |
… | |
… | |
279 | inline void |
284 | inline void |
280 | encode_comma (enc_t *enc) |
285 | encode_comma (enc_t *enc) |
281 | { |
286 | { |
282 | encode_ch (enc, ','); |
287 | encode_ch (enc, ','); |
283 | |
288 | |
284 | if (enc->flags & F_INDENT) |
289 | if (enc->json.flags & F_INDENT) |
285 | encode_nl (enc); |
290 | encode_nl (enc); |
286 | else if (enc->flags & F_SPACE_AFTER) |
291 | else if (enc->json.flags & F_SPACE_AFTER) |
287 | encode_space (enc); |
292 | encode_space (enc); |
288 | } |
293 | } |
289 | |
294 | |
290 | static void encode_sv (enc_t *enc, SV *sv); |
295 | static void encode_sv (enc_t *enc, SV *sv); |
291 | |
296 | |
… | |
… | |
334 | else |
339 | else |
335 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
340 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
336 | |
341 | |
337 | encode_ch (enc, '"'); |
342 | encode_ch (enc, '"'); |
338 | |
343 | |
339 | if (enc->flags & F_SPACE_BEFORE) encode_space (enc); |
344 | if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); |
340 | encode_ch (enc, ':'); |
345 | encode_ch (enc, ':'); |
341 | if (enc->flags & F_SPACE_AFTER ) encode_space (enc); |
346 | if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); |
342 | encode_sv (enc, HeVAL (he)); |
347 | encode_sv (enc, HeVAL (he)); |
343 | } |
348 | } |
344 | |
349 | |
345 | // compare hash entries, used when all keys are bytestrings |
350 | // compare hash entries, used when all keys are bytestrings |
346 | static int |
351 | static int |
… | |
… | |
381 | { |
386 | { |
382 | // for canonical output we have to sort by keys first |
387 | // for canonical output we have to sort by keys first |
383 | // actually, this is mostly due to the stupid so-called |
388 | // actually, this is mostly due to the stupid so-called |
384 | // security workaround added somewhere in 5.8.x. |
389 | // security workaround added somewhere in 5.8.x. |
385 | // that randomises hash orderings |
390 | // that randomises hash orderings |
386 | if (enc->flags & F_CANONICAL) |
391 | if (enc->json.flags & F_CANONICAL) |
387 | { |
392 | { |
388 | int fast = 1; |
393 | int fast = 1; |
389 | HE *he; |
394 | HE *he; |
390 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
395 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
391 | HE **hes = _alloca (count * sizeof (HE)); |
396 | HE **hes = _alloca (count * sizeof (HE)); |
… | |
… | |
463 | svtype svt; |
468 | svtype svt; |
464 | |
469 | |
465 | SvGETMAGIC (sv); |
470 | SvGETMAGIC (sv); |
466 | svt = SvTYPE (sv); |
471 | svt = SvTYPE (sv); |
467 | |
472 | |
|
|
473 | if (expect_false (SvOBJECT (sv))) |
|
|
474 | { |
|
|
475 | if (SvSTASH (sv) == json_boolean_stash) |
|
|
476 | { |
|
|
477 | if (SvIV (sv) == 0) |
|
|
478 | encode_str (enc, "false", 5, 0); |
|
|
479 | else |
|
|
480 | encode_str (enc, "true", 4, 0); |
|
|
481 | } |
|
|
482 | else |
|
|
483 | { |
|
|
484 | #if 0 |
|
|
485 | if (0 && sv_derived_from (rv, "JSON::Literal")) |
|
|
486 | { |
|
|
487 | // not yet |
|
|
488 | } |
|
|
489 | #endif |
|
|
490 | if (enc->json.flags & F_CONV_BLESSED) |
|
|
491 | { |
|
|
492 | // we re-bless the reference to get overload and other niceties right |
|
|
493 | GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1); |
|
|
494 | |
|
|
495 | if (to_json) |
|
|
496 | { |
|
|
497 | dSP; ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
498 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
|
|
499 | |
|
|
500 | // calling with G_SCALAR ensures that we always get a 1 reutrn value |
|
|
501 | // check anyways. |
|
|
502 | PUTBACK; |
|
|
503 | assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR)); |
|
|
504 | SPAGAIN; |
|
|
505 | |
|
|
506 | encode_sv (enc, POPs); |
|
|
507 | |
|
|
508 | FREETMPS; LEAVE; |
|
|
509 | } |
|
|
510 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
511 | encode_str (enc, "null", 4, 0); |
|
|
512 | else |
|
|
513 | croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", |
|
|
514 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
515 | } |
|
|
516 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
517 | encode_str (enc, "null", 4, 0); |
|
|
518 | else |
|
|
519 | croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", |
|
|
520 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
521 | } |
|
|
522 | } |
468 | if (svt == SVt_PVHV) |
523 | else if (svt == SVt_PVHV) |
469 | encode_hv (enc, (HV *)sv); |
524 | encode_hv (enc, (HV *)sv); |
470 | else if (svt == SVt_PVAV) |
525 | else if (svt == SVt_PVAV) |
471 | encode_av (enc, (AV *)sv); |
526 | encode_av (enc, (AV *)sv); |
472 | else if (svt < SVt_PVAV) |
527 | else if (svt < SVt_PVAV) |
473 | { |
528 | { |
… | |
… | |
552 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
607 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
553 | SvPV_nolen (sv), SvFLAGS (sv)); |
608 | SvPV_nolen (sv), SvFLAGS (sv)); |
554 | } |
609 | } |
555 | |
610 | |
556 | static SV * |
611 | static SV * |
557 | encode_json (SV *scalar, U32 flags) |
612 | encode_json (SV *scalar, JSON *json) |
558 | { |
613 | { |
559 | enc_t enc; |
614 | enc_t enc; |
560 | |
615 | |
561 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
616 | if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
562 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
617 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
563 | |
618 | |
564 | enc.flags = flags; |
619 | enc.json = *json; |
565 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
620 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
566 | enc.cur = SvPVX (enc.sv); |
621 | enc.cur = SvPVX (enc.sv); |
567 | enc.end = SvEND (enc.sv); |
622 | enc.end = SvEND (enc.sv); |
568 | enc.indent = 0; |
623 | enc.indent = 0; |
569 | enc.maxdepth = DEC_DEPTH (flags); |
624 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
570 | |
625 | |
571 | SvPOK_only (enc.sv); |
626 | SvPOK_only (enc.sv); |
572 | encode_sv (&enc, scalar); |
627 | encode_sv (&enc, scalar); |
573 | |
628 | |
574 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
629 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
575 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
630 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
576 | |
631 | |
577 | if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
632 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
578 | SvUTF8_on (enc.sv); |
633 | SvUTF8_on (enc.sv); |
579 | |
634 | |
580 | if (enc.flags & F_SHRINK) |
635 | if (enc.json.flags & F_SHRINK) |
581 | shrink (enc.sv); |
636 | shrink (enc.sv); |
582 | |
637 | |
583 | return enc.sv; |
638 | return enc.sv; |
584 | } |
639 | } |
585 | |
640 | |
… | |
… | |
590 | typedef struct |
645 | typedef struct |
591 | { |
646 | { |
592 | char *cur; // current parser pointer |
647 | char *cur; // current parser pointer |
593 | char *end; // end of input string |
648 | char *end; // end of input string |
594 | const char *err; // parse error, if != 0 |
649 | const char *err; // parse error, if != 0 |
595 | U32 flags; // F_* |
650 | JSON json; |
596 | U32 depth; // recursion depth |
651 | U32 depth; // recursion depth |
597 | U32 maxdepth; // recursion depth limit |
652 | U32 maxdepth; // recursion depth limit |
598 | } dec_t; |
653 | } dec_t; |
599 | |
654 | |
600 | inline void |
655 | inline void |
… | |
… | |
631 | decode_4hex (dec_t *dec) |
686 | decode_4hex (dec_t *dec) |
632 | { |
687 | { |
633 | signed char d1, d2, d3, d4; |
688 | signed char d1, d2, d3, d4; |
634 | unsigned char *cur = (unsigned char *)dec->cur; |
689 | unsigned char *cur = (unsigned char *)dec->cur; |
635 | |
690 | |
636 | d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected"); |
691 | d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected"); |
637 | d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected"); |
692 | d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected"); |
638 | d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected"); |
693 | d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected"); |
639 | d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected"); |
694 | d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected"); |
640 | |
695 | |
641 | dec->cur += 4; |
696 | dec->cur += 4; |
642 | |
697 | |
643 | return ((UV)d1) << 12 |
698 | return ((UV)d1) << 12 |
644 | | ((UV)d2) << 8 |
699 | | ((UV)d2) << 8 |
… | |
… | |
867 | { |
922 | { |
868 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
923 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
869 | if (*start == '-') |
924 | if (*start == '-') |
870 | switch (dec->cur - start) |
925 | switch (dec->cur - start) |
871 | { |
926 | { |
872 | case 2: return newSViv (-( start [1] - '0' )); |
927 | case 2: return newSViv (-( start [1] - '0' * 1)); |
873 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
928 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
874 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
929 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
875 | case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
930 | case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
876 | } |
931 | } |
877 | else |
932 | else |
878 | switch (dec->cur - start) |
933 | switch (dec->cur - start) |
879 | { |
934 | { |
880 | case 1: return newSViv ( start [0] - '0' ); |
935 | case 1: return newSViv ( start [0] - '0' * 1); |
881 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
936 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
882 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
937 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
883 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
938 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
884 | } |
939 | } |
885 | |
940 | |
… | |
… | |
892 | if (uv < (UV)IV_MIN) |
947 | if (uv < (UV)IV_MIN) |
893 | return newSViv (-(IV)uv); |
948 | return newSViv (-(IV)uv); |
894 | } |
949 | } |
895 | else |
950 | else |
896 | return newSVuv (uv); |
951 | return newSVuv (uv); |
|
|
952 | |
|
|
953 | // here would likely be the place for bigint support |
897 | } |
954 | } |
898 | } |
955 | } |
899 | |
956 | |
|
|
957 | // if we ever support bigint or bigfloat, this is the place for bigfloat |
900 | return newSVnv (Atof (start)); |
958 | return newSVnv (Atof (start)); |
901 | |
959 | |
902 | fail: |
960 | fail: |
903 | return 0; |
961 | return 0; |
904 | } |
962 | } |
… | |
… | |
948 | } |
1006 | } |
949 | |
1007 | |
950 | static SV * |
1008 | static SV * |
951 | decode_hv (dec_t *dec) |
1009 | decode_hv (dec_t *dec) |
952 | { |
1010 | { |
|
|
1011 | SV *sv; |
953 | HV *hv = newHV (); |
1012 | HV *hv = newHV (); |
954 | |
1013 | |
955 | DEC_INC_DEPTH; |
1014 | DEC_INC_DEPTH; |
956 | decode_ws (dec); |
1015 | decode_ws (dec); |
957 | |
1016 | |
958 | if (*dec->cur == '}') |
1017 | if (*dec->cur == '}') |
959 | ++dec->cur; |
1018 | ++dec->cur; |
960 | else |
1019 | else |
961 | for (;;) |
1020 | for (;;) |
962 | { |
1021 | { |
963 | SV *key, *value; |
|
|
964 | |
|
|
965 | decode_ws (dec); EXPECT_CH ('"'); |
1022 | decode_ws (dec); EXPECT_CH ('"'); |
966 | |
1023 | |
967 | key = decode_str (dec); |
1024 | // heuristic: assume that |
968 | if (!key) |
1025 | // a) decode_str + hv_store_ent are abysmally slow. |
969 | goto fail; |
1026 | // b) most hash keys are short, simple ascii text. |
|
|
1027 | // => try to "fast-match" such strings to avoid |
|
|
1028 | // the overhead of decode_str + hv_store_ent. |
|
|
1029 | { |
|
|
1030 | SV *value; |
|
|
1031 | char *p = dec->cur; |
|
|
1032 | char *e = p + 24; // only try up to 24 bytes |
970 | |
1033 | |
971 | decode_ws (dec); EXPECT_CH (':'); |
1034 | for (;;) |
972 | |
|
|
973 | value = decode_sv (dec); |
|
|
974 | if (!value) |
|
|
975 | { |
1035 | { |
|
|
1036 | // the >= 0x80 is true on most architectures |
|
|
1037 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
|
|
1038 | { |
|
|
1039 | // slow path, back up and use decode_str |
|
|
1040 | SV *key = decode_str (dec); |
|
|
1041 | if (!key) |
|
|
1042 | goto fail; |
|
|
1043 | |
|
|
1044 | decode_ws (dec); EXPECT_CH (':'); |
|
|
1045 | |
|
|
1046 | value = decode_sv (dec); |
|
|
1047 | if (!value) |
|
|
1048 | { |
|
|
1049 | SvREFCNT_dec (key); |
|
|
1050 | goto fail; |
|
|
1051 | } |
|
|
1052 | |
|
|
1053 | hv_store_ent (hv, key, value, 0); |
976 | SvREFCNT_dec (key); |
1054 | SvREFCNT_dec (key); |
|
|
1055 | |
|
|
1056 | break; |
|
|
1057 | } |
|
|
1058 | else if (*p == '"') |
|
|
1059 | { |
|
|
1060 | // fast path, got a simple key |
|
|
1061 | char *key = dec->cur; |
|
|
1062 | int len = p - key; |
|
|
1063 | dec->cur = p + 1; |
|
|
1064 | |
|
|
1065 | decode_ws (dec); EXPECT_CH (':'); |
|
|
1066 | |
|
|
1067 | value = decode_sv (dec); |
|
|
1068 | if (!value) |
977 | goto fail; |
1069 | goto fail; |
|
|
1070 | |
|
|
1071 | hv_store (hv, key, len, value, 0); |
|
|
1072 | |
|
|
1073 | break; |
|
|
1074 | } |
|
|
1075 | |
|
|
1076 | ++p; |
978 | } |
1077 | } |
979 | |
1078 | } |
980 | hv_store_ent (hv, key, value, 0); |
|
|
981 | SvREFCNT_dec (key); |
|
|
982 | |
1079 | |
983 | decode_ws (dec); |
1080 | decode_ws (dec); |
984 | |
1081 | |
985 | if (*dec->cur == '}') |
1082 | if (*dec->cur == '}') |
986 | { |
1083 | { |
… | |
… | |
993 | |
1090 | |
994 | ++dec->cur; |
1091 | ++dec->cur; |
995 | } |
1092 | } |
996 | |
1093 | |
997 | DEC_DEC_DEPTH; |
1094 | DEC_DEC_DEPTH; |
|
|
1095 | sv = newRV_noinc ((SV *)hv); |
|
|
1096 | |
|
|
1097 | // check filter callbacks |
|
|
1098 | if (dec->json.flags & F_HOOK) |
|
|
1099 | { |
|
|
1100 | ENTER; SAVETMPS; |
|
|
1101 | |
|
|
1102 | if (HvKEYS (hv) == 1 && dec->json.cb_sk_object) |
|
|
1103 | { |
|
|
1104 | int count; |
|
|
1105 | |
|
|
1106 | dSP; PUSHMARK (SP); |
|
|
1107 | XPUSHs (sv_2mortal (sv)); |
|
|
1108 | |
|
|
1109 | PUTBACK; count = call_sv (dec->json.cb_sk_object, G_ARRAY); SPAGAIN; |
|
|
1110 | |
|
|
1111 | if (count == 1) |
|
|
1112 | sv = newSVsv (POPs); |
|
|
1113 | else |
|
|
1114 | SvREFCNT_inc (sv); |
|
|
1115 | } |
|
|
1116 | |
|
|
1117 | if (dec->json.cb_object) |
|
|
1118 | { |
|
|
1119 | int count; |
|
|
1120 | |
|
|
1121 | dSP; ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
1122 | XPUSHs (sv_2mortal (sv)); |
|
|
1123 | |
|
|
1124 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
|
|
1125 | |
|
|
1126 | if (count == 1) |
|
|
1127 | sv = newSVsv (POPs); |
|
|
1128 | else |
|
|
1129 | SvREFCNT_inc (sv); |
|
|
1130 | } |
|
|
1131 | |
|
|
1132 | FREETMPS; LEAVE; |
|
|
1133 | } |
|
|
1134 | |
998 | return newRV_noinc ((SV *)hv); |
1135 | return newRV_noinc ((SV *)hv); |
999 | |
1136 | |
1000 | fail: |
1137 | fail: |
1001 | SvREFCNT_dec (hv); |
1138 | SvREFCNT_dec (hv); |
1002 | DEC_DEC_DEPTH; |
1139 | DEC_DEC_DEPTH; |
… | |
… | |
1005 | |
1142 | |
1006 | static SV * |
1143 | static SV * |
1007 | decode_sv (dec_t *dec) |
1144 | decode_sv (dec_t *dec) |
1008 | { |
1145 | { |
1009 | decode_ws (dec); |
1146 | decode_ws (dec); |
|
|
1147 | |
|
|
1148 | // the beauty of JSON: you need exactly one character lookahead |
|
|
1149 | // to parse anything. |
1010 | switch (*dec->cur) |
1150 | switch (*dec->cur) |
1011 | { |
1151 | { |
1012 | case '"': ++dec->cur; return decode_str (dec); |
1152 | case '"': ++dec->cur; return decode_str (dec); |
1013 | case '[': ++dec->cur; return decode_av (dec); |
1153 | case '[': ++dec->cur; return decode_av (dec); |
1014 | case '{': ++dec->cur; return decode_hv (dec); |
1154 | case '{': ++dec->cur; return decode_hv (dec); |
… | |
… | |
1020 | |
1160 | |
1021 | case 't': |
1161 | case 't': |
1022 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1162 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1023 | { |
1163 | { |
1024 | dec->cur += 4; |
1164 | dec->cur += 4; |
1025 | return newSViv (1); |
1165 | return SvREFCNT_inc (json_true); |
1026 | } |
1166 | } |
1027 | else |
1167 | else |
1028 | ERR ("'true' expected"); |
1168 | ERR ("'true' expected"); |
1029 | |
1169 | |
1030 | break; |
1170 | break; |
1031 | |
1171 | |
1032 | case 'f': |
1172 | case 'f': |
1033 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1173 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1034 | { |
1174 | { |
1035 | dec->cur += 5; |
1175 | dec->cur += 5; |
1036 | return newSViv (0); |
1176 | return SvREFCNT_inc (json_false); |
1037 | } |
1177 | } |
1038 | else |
1178 | else |
1039 | ERR ("'false' expected"); |
1179 | ERR ("'false' expected"); |
1040 | |
1180 | |
1041 | break; |
1181 | break; |
… | |
… | |
1059 | fail: |
1199 | fail: |
1060 | return 0; |
1200 | return 0; |
1061 | } |
1201 | } |
1062 | |
1202 | |
1063 | static SV * |
1203 | static SV * |
1064 | decode_json (SV *string, U32 flags, UV *offset_return) |
1204 | decode_json (SV *string, JSON *json, UV *offset_return) |
1065 | { |
1205 | { |
1066 | dec_t dec; |
1206 | dec_t dec; |
1067 | UV offset; |
1207 | UV offset; |
1068 | SV *sv; |
1208 | SV *sv; |
1069 | |
1209 | |
1070 | SvGETMAGIC (string); |
1210 | SvGETMAGIC (string); |
1071 | SvUPGRADE (string, SVt_PV); |
1211 | SvUPGRADE (string, SVt_PV); |
1072 | |
1212 | |
|
|
1213 | if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) |
|
|
1214 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
1215 | (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); |
|
|
1216 | |
1073 | if (flags & F_UTF8) |
1217 | if (json->flags & F_UTF8) |
1074 | sv_utf8_downgrade (string, 0); |
1218 | sv_utf8_downgrade (string, 0); |
1075 | else |
1219 | else |
1076 | sv_utf8_upgrade (string); |
1220 | sv_utf8_upgrade (string); |
1077 | |
1221 | |
1078 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1222 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1079 | |
1223 | |
1080 | dec.flags = flags; |
1224 | dec.json = *json; |
1081 | dec.cur = SvPVX (string); |
1225 | dec.cur = SvPVX (string); |
1082 | dec.end = SvEND (string); |
1226 | dec.end = SvEND (string); |
1083 | dec.err = 0; |
1227 | dec.err = 0; |
1084 | dec.depth = 0; |
1228 | dec.depth = 0; |
1085 | dec.maxdepth = DEC_DEPTH (dec.flags); |
1229 | dec.maxdepth = DEC_DEPTH (dec.json.flags); |
|
|
1230 | |
|
|
1231 | if (dec.json.cb_object || dec.json.cb_sk_object) |
|
|
1232 | dec.json.flags |= F_HOOK; |
1086 | |
1233 | |
1087 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1234 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1088 | sv = decode_sv (&dec); |
1235 | sv = decode_sv (&dec); |
1089 | |
1236 | |
1090 | if (!(offset_return || !sv)) |
1237 | if (!(offset_return || !sv)) |
… | |
… | |
1100 | } |
1247 | } |
1101 | } |
1248 | } |
1102 | |
1249 | |
1103 | if (offset_return || !sv) |
1250 | if (offset_return || !sv) |
1104 | { |
1251 | { |
1105 | offset = dec.flags & F_UTF8 |
1252 | offset = dec.json.flags & F_UTF8 |
1106 | ? dec.cur - SvPVX (string) |
1253 | ? dec.cur - SvPVX (string) |
1107 | : utf8_distance (dec.cur, SvPVX (string)); |
1254 | : utf8_distance (dec.cur, SvPVX (string)); |
1108 | |
1255 | |
1109 | if (offset_return) |
1256 | if (offset_return) |
1110 | *offset_return = offset; |
1257 | *offset_return = offset; |
… | |
… | |
1129 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1276 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1130 | } |
1277 | } |
1131 | |
1278 | |
1132 | sv = sv_2mortal (sv); |
1279 | sv = sv_2mortal (sv); |
1133 | |
1280 | |
1134 | if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1281 | if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1135 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1282 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1136 | |
1283 | |
1137 | return sv; |
1284 | return sv; |
1138 | } |
1285 | } |
1139 | |
1286 | |
… | |
… | |
1143 | MODULE = JSON::XS PACKAGE = JSON::XS |
1290 | MODULE = JSON::XS PACKAGE = JSON::XS |
1144 | |
1291 | |
1145 | BOOT: |
1292 | BOOT: |
1146 | { |
1293 | { |
1147 | int i; |
1294 | int i; |
1148 | |
|
|
1149 | memset (decode_hexdigit, 0xff, 256); |
|
|
1150 | |
1295 | |
1151 | for (i = 0; i < 256; ++i) |
1296 | for (i = 0; i < 256; ++i) |
1152 | decode_hexdigit [i] = |
1297 | decode_hexdigit [i] = |
1153 | i >= '0' && i <= '9' ? i - '0' |
1298 | i >= '0' && i <= '9' ? i - '0' |
1154 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1299 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1155 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1300 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1156 | : -1; |
1301 | : -1; |
1157 | |
1302 | |
1158 | json_stash = gv_stashpv ("JSON::XS", 1); |
1303 | json_stash = gv_stashpv ("JSON::XS" , 1); |
|
|
1304 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
1305 | |
|
|
1306 | json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
|
|
1307 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1159 | } |
1308 | } |
1160 | |
1309 | |
1161 | PROTOTYPES: DISABLE |
1310 | PROTOTYPES: DISABLE |
1162 | |
1311 | |
1163 | SV *new (char *dummy) |
1312 | void new (char *klass) |
1164 | CODE: |
1313 | PPCODE: |
1165 | RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); |
1314 | { |
1166 | OUTPUT: |
1315 | SV *pv = NEWSV (0, sizeof (JSON)); |
1167 | RETVAL |
1316 | SvPOK_only (pv); |
|
|
1317 | Zero (SvPVX (pv), 1, JSON); |
|
|
1318 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
|
|
1319 | XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash))); |
|
|
1320 | } |
1168 | |
1321 | |
1169 | SV *ascii (SV *self, int enable = 1) |
1322 | void ascii (JSON *self, int enable = 1) |
1170 | ALIAS: |
1323 | ALIAS: |
1171 | ascii = F_ASCII |
1324 | ascii = F_ASCII |
1172 | latin1 = F_LATIN1 |
1325 | latin1 = F_LATIN1 |
1173 | utf8 = F_UTF8 |
1326 | utf8 = F_UTF8 |
1174 | indent = F_INDENT |
1327 | indent = F_INDENT |
1175 | canonical = F_CANONICAL |
1328 | canonical = F_CANONICAL |
1176 | space_before = F_SPACE_BEFORE |
1329 | space_before = F_SPACE_BEFORE |
1177 | space_after = F_SPACE_AFTER |
1330 | space_after = F_SPACE_AFTER |
1178 | pretty = F_PRETTY |
1331 | pretty = F_PRETTY |
1179 | allow_nonref = F_ALLOW_NONREF |
1332 | allow_nonref = F_ALLOW_NONREF |
1180 | shrink = F_SHRINK |
1333 | shrink = F_SHRINK |
|
|
1334 | allow_blessed = F_ALLOW_BLESSED |
|
|
1335 | convert_blessed = F_CONV_BLESSED |
1181 | CODE: |
1336 | PPCODE: |
1182 | { |
1337 | { |
1183 | UV *uv = SvJSON (self); |
|
|
1184 | if (enable) |
1338 | if (enable) |
1185 | *uv |= ix; |
1339 | self->flags |= ix; |
1186 | else |
1340 | else |
1187 | *uv &= ~ix; |
1341 | self->flags &= ~ix; |
1188 | |
1342 | |
1189 | RETVAL = newSVsv (self); |
1343 | XPUSHs (ST (0)); |
1190 | } |
1344 | } |
1191 | OUTPUT: |
|
|
1192 | RETVAL |
|
|
1193 | |
1345 | |
1194 | SV *max_depth (SV *self, UV max_depth = 0x80000000UL) |
1346 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
1195 | CODE: |
1347 | PPCODE: |
1196 | { |
1348 | { |
1197 | UV *uv = SvJSON (self); |
|
|
1198 | UV log2 = 0; |
1349 | UV log2 = 0; |
1199 | |
1350 | |
1200 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
1351 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
1201 | |
1352 | |
1202 | while ((1UL << log2) < max_depth) |
1353 | while ((1UL << log2) < max_depth) |
1203 | ++log2; |
1354 | ++log2; |
1204 | |
1355 | |
1205 | *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
1356 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
1206 | |
1357 | |
1207 | RETVAL = newSVsv (self); |
1358 | XPUSHs (ST (0)); |
1208 | } |
1359 | } |
1209 | OUTPUT: |
|
|
1210 | RETVAL |
|
|
1211 | |
1360 | |
1212 | void encode (SV *self, SV *scalar) |
1361 | void max_size (JSON *self, UV max_size = 0) |
1213 | PPCODE: |
1362 | PPCODE: |
1214 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1363 | { |
|
|
1364 | UV log2 = 0; |
1215 | |
1365 | |
1216 | void decode (SV *self, SV *jsonstr) |
1366 | if (max_size > 0x80000000UL) max_size = 0x80000000UL; |
|
|
1367 | if (max_size == 1) max_size = 2; |
|
|
1368 | |
|
|
1369 | while ((1UL << log2) < max_size) |
|
|
1370 | ++log2; |
|
|
1371 | |
|
|
1372 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
|
|
1373 | |
|
|
1374 | XPUSHs (ST (0)); |
|
|
1375 | } |
|
|
1376 | |
|
|
1377 | void filter_json_objects (JSON *self, SV *cb = &PL_sv_undef) |
|
|
1378 | ALIAS: |
|
|
1379 | filter_sk_json_objects = 1 |
1217 | PPCODE: |
1380 | PPCODE: |
|
|
1381 | { |
|
|
1382 | if (!SvOK (cb)) |
|
|
1383 | cb = 0; |
|
|
1384 | |
|
|
1385 | switch (ix) |
|
|
1386 | { |
|
|
1387 | case 0: self->cb_object = cb; break; |
|
|
1388 | case 1: self->cb_sk_object = cb; break; |
|
|
1389 | } |
|
|
1390 | |
|
|
1391 | XPUSHs (ST (0)); |
|
|
1392 | } |
|
|
1393 | |
|
|
1394 | void encode (JSON *self, SV *scalar) |
|
|
1395 | PPCODE: |
|
|
1396 | XPUSHs (encode_json (scalar, self)); |
|
|
1397 | |
|
|
1398 | void decode (JSON *self, SV *jsonstr) |
|
|
1399 | PPCODE: |
1218 | XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); |
1400 | XPUSHs (decode_json (jsonstr, self, 0)); |
1219 | |
1401 | |
1220 | void decode_prefix (SV *self, SV *jsonstr) |
1402 | void decode_prefix (JSON *self, SV *jsonstr) |
1221 | PPCODE: |
1403 | PPCODE: |
1222 | { |
1404 | { |
1223 | UV offset; |
1405 | UV offset; |
1224 | EXTEND (SP, 2); |
1406 | EXTEND (SP, 2); |
1225 | PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); |
1407 | PUSHs (decode_json (jsonstr, self, &offset)); |
1226 | PUSHs (sv_2mortal (newSVuv (offset))); |
1408 | PUSHs (sv_2mortal (newSVuv (offset))); |
1227 | } |
1409 | } |
1228 | |
1410 | |
1229 | PROTOTYPES: ENABLE |
1411 | PROTOTYPES: ENABLE |
1230 | |
1412 | |
1231 | void to_json (SV *scalar) |
1413 | void to_json (SV *scalar) |
1232 | ALIAS: |
|
|
1233 | objToJson = 0 |
|
|
1234 | PPCODE: |
1414 | PPCODE: |
|
|
1415 | { |
|
|
1416 | JSON json = { F_DEFAULT | F_UTF8 }; |
1235 | XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); |
1417 | XPUSHs (encode_json (scalar, &json)); |
|
|
1418 | } |
1236 | |
1419 | |
1237 | void from_json (SV *jsonstr) |
1420 | void from_json (SV *jsonstr) |
1238 | ALIAS: |
|
|
1239 | jsonToObj = 0 |
|
|
1240 | PPCODE: |
1421 | PPCODE: |
|
|
1422 | { |
|
|
1423 | JSON json = { F_DEFAULT | F_UTF8 }; |
1241 | XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); |
1424 | XPUSHs (decode_json (jsonstr, &json, 0)); |
|
|
1425 | } |
1242 | |
1426 | |