1 | #include "EXTERN.h" |
1 | #include "EXTERN.h" |
2 | #include "perl.h" |
2 | #include "perl.h" |
3 | #include "XSUB.h" |
3 | #include "XSUB.h" |
4 | |
4 | |
5 | #include "assert.h" |
5 | #include <assert.h> |
6 | #include "string.h" |
6 | #include <string.h> |
7 | #include "stdlib.h" |
7 | #include <stdlib.h> |
8 | #include "stdio.h" |
8 | #include <stdio.h> |
|
|
9 | #include <float.h> |
9 | |
10 | |
10 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
11 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
11 | # define snprintf _snprintf // C compilers have this in stdio.h |
12 | # define snprintf _snprintf // C compilers have this in stdio.h |
12 | #endif |
13 | #endif |
13 | |
14 | |
|
|
15 | // some old perls do not have this, try to make it work, no |
|
|
16 | // guarentees, though. if it breaks, you get to keep the pieces. |
|
|
17 | #ifndef UTF8_MAXBYTES |
|
|
18 | # define UTF8_MAXBYTES 13 |
|
|
19 | #endif |
|
|
20 | |
14 | #define F_ASCII 0x00000001UL |
21 | #define F_ASCII 0x00000001UL |
15 | #define F_LATIN1 0x00000002UL |
22 | #define F_LATIN1 0x00000002UL |
16 | #define F_UTF8 0x00000004UL |
23 | #define F_UTF8 0x00000004UL |
17 | #define F_INDENT 0x00000008UL |
24 | #define F_INDENT 0x00000008UL |
18 | #define F_CANONICAL 0x00000010UL |
25 | #define F_CANONICAL 0x00000010UL |
19 | #define F_SPACE_BEFORE 0x00000020UL |
26 | #define F_SPACE_BEFORE 0x00000020UL |
20 | #define F_SPACE_AFTER 0x00000040UL |
27 | #define F_SPACE_AFTER 0x00000040UL |
21 | #define F_ALLOW_NONREF 0x00000100UL |
28 | #define F_ALLOW_NONREF 0x00000100UL |
22 | #define F_SHRINK 0x00000200UL |
29 | #define F_SHRINK 0x00000200UL |
|
|
30 | #define F_ALLOW_BLESSED 0x00000400UL |
|
|
31 | #define F_CONV_BLESSED 0x00000800UL |
23 | #define F_MAXDEPTH 0xf8000000UL |
32 | #define F_MAXDEPTH 0xf8000000UL |
24 | #define S_MAXDEPTH 27 |
33 | #define S_MAXDEPTH 27 |
|
|
34 | #define F_MAXSIZE 0x01f00000UL |
|
|
35 | #define S_MAXSIZE 20 |
|
|
36 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
25 | |
37 | |
26 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
38 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
27 | |
39 | #define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE )) |
28 | // F_SELFCONVERT? <=> to_json/toJson |
|
|
29 | // F_BLESSED? <=> { $__class__$ => } |
|
|
30 | |
40 | |
31 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
41 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
32 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
42 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
33 | |
43 | |
34 | #define INIT_SIZE 32 // initial scalar size to be allocated |
44 | #define INIT_SIZE 32 // initial scalar size to be allocated |
… | |
… | |
48 | #endif |
58 | #endif |
49 | |
59 | |
50 | #define expect_false(expr) expect ((expr) != 0, 0) |
60 | #define expect_false(expr) expect ((expr) != 0, 0) |
51 | #define expect_true(expr) expect ((expr) != 0, 1) |
61 | #define expect_true(expr) expect ((expr) != 0, 1) |
52 | |
62 | |
|
|
63 | #ifdef USE_ITHREADS |
|
|
64 | # define JSON_SLOW 1 |
|
|
65 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
|
|
66 | #else |
|
|
67 | # define JSON_SLOW 0 |
|
|
68 | # define JSON_STASH json_stash |
|
|
69 | #endif |
|
|
70 | |
53 | static HV *json_stash; // JSON::XS:: |
71 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
|
|
72 | static SV *json_true, *json_false; |
|
|
73 | |
|
|
74 | typedef struct { |
|
|
75 | U32 flags; |
|
|
76 | SV *cb_object; |
|
|
77 | HV *cb_sk_object; |
|
|
78 | } JSON; |
54 | |
79 | |
55 | ///////////////////////////////////////////////////////////////////////////// |
80 | ///////////////////////////////////////////////////////////////////////////// |
56 | // utility functions |
81 | // utility functions |
57 | |
82 | |
58 | static UV * |
83 | 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) |
84 | shrink (SV *sv) |
69 | { |
85 | { |
70 | sv_utf8_downgrade (sv, 1); |
86 | sv_utf8_downgrade (sv, 1); |
71 | if (SvLEN (sv) > SvCUR (sv) + 1) |
87 | if (SvLEN (sv) > SvCUR (sv) + 1) |
72 | { |
88 | { |
… | |
… | |
107 | typedef struct |
123 | typedef struct |
108 | { |
124 | { |
109 | char *cur; // SvPVX (sv) + current output position |
125 | char *cur; // SvPVX (sv) + current output position |
110 | char *end; // SvEND (sv) |
126 | char *end; // SvEND (sv) |
111 | SV *sv; // result scalar |
127 | SV *sv; // result scalar |
112 | U32 flags; // F_* |
128 | JSON json; |
113 | U32 indent; // indentation level |
129 | U32 indent; // indentation level |
114 | U32 maxdepth; // max. indentation/recursion level |
130 | U32 maxdepth; // max. indentation/recursion level |
115 | } enc_t; |
131 | } enc_t; |
116 | |
132 | |
117 | inline void |
133 | inline void |
… | |
… | |
178 | STRLEN clen; |
194 | STRLEN clen; |
179 | UV uch; |
195 | UV uch; |
180 | |
196 | |
181 | if (is_utf8) |
197 | if (is_utf8) |
182 | { |
198 | { |
183 | //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
|
|
184 | uch = decode_utf8 (str, end - str, &clen); |
199 | uch = decode_utf8 (str, end - str, &clen); |
185 | if (clen == (STRLEN)-1) |
200 | if (clen == (STRLEN)-1) |
186 | croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
201 | croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
187 | } |
202 | } |
188 | else |
203 | else |
… | |
… | |
192 | } |
207 | } |
193 | |
208 | |
194 | if (uch > 0x10FFFFUL) |
209 | if (uch > 0x10FFFFUL) |
195 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
210 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
196 | |
211 | |
197 | if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF)) |
212 | if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF)) |
198 | { |
213 | { |
199 | if (uch > 0xFFFFUL) |
214 | if (uch > 0xFFFFUL) |
200 | { |
215 | { |
201 | need (enc, len += 11); |
216 | need (enc, len += 11); |
202 | sprintf (enc->cur, "\\u%04x\\u%04x", |
217 | sprintf (enc->cur, "\\u%04x\\u%04x", |
… | |
… | |
216 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
231 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
217 | } |
232 | } |
218 | |
233 | |
219 | str += clen; |
234 | str += clen; |
220 | } |
235 | } |
221 | else if (enc->flags & F_LATIN1) |
236 | else if (enc->json.flags & F_LATIN1) |
222 | { |
237 | { |
223 | *enc->cur++ = uch; |
238 | *enc->cur++ = uch; |
224 | str += clen; |
239 | str += clen; |
225 | } |
240 | } |
226 | else if (is_utf8) |
241 | else if (is_utf8) |
… | |
… | |
247 | } |
262 | } |
248 | |
263 | |
249 | inline void |
264 | inline void |
250 | encode_indent (enc_t *enc) |
265 | encode_indent (enc_t *enc) |
251 | { |
266 | { |
252 | if (enc->flags & F_INDENT) |
267 | if (enc->json.flags & F_INDENT) |
253 | { |
268 | { |
254 | int spaces = enc->indent * INDENT_STEP; |
269 | int spaces = enc->indent * INDENT_STEP; |
255 | |
270 | |
256 | need (enc, spaces); |
271 | need (enc, spaces); |
257 | memset (enc->cur, ' ', spaces); |
272 | memset (enc->cur, ' ', spaces); |
… | |
… | |
267 | } |
282 | } |
268 | |
283 | |
269 | inline void |
284 | inline void |
270 | encode_nl (enc_t *enc) |
285 | encode_nl (enc_t *enc) |
271 | { |
286 | { |
272 | if (enc->flags & F_INDENT) |
287 | if (enc->json.flags & F_INDENT) |
273 | { |
288 | { |
274 | need (enc, 1); |
289 | need (enc, 1); |
275 | encode_ch (enc, '\n'); |
290 | encode_ch (enc, '\n'); |
276 | } |
291 | } |
277 | } |
292 | } |
… | |
… | |
279 | inline void |
294 | inline void |
280 | encode_comma (enc_t *enc) |
295 | encode_comma (enc_t *enc) |
281 | { |
296 | { |
282 | encode_ch (enc, ','); |
297 | encode_ch (enc, ','); |
283 | |
298 | |
284 | if (enc->flags & F_INDENT) |
299 | if (enc->json.flags & F_INDENT) |
285 | encode_nl (enc); |
300 | encode_nl (enc); |
286 | else if (enc->flags & F_SPACE_AFTER) |
301 | else if (enc->json.flags & F_SPACE_AFTER) |
287 | encode_space (enc); |
302 | encode_space (enc); |
288 | } |
303 | } |
289 | |
304 | |
290 | static void encode_sv (enc_t *enc, SV *sv); |
305 | static void encode_sv (enc_t *enc, SV *sv); |
291 | |
306 | |
… | |
… | |
300 | encode_ch (enc, '['); encode_nl (enc); |
315 | encode_ch (enc, '['); encode_nl (enc); |
301 | ++enc->indent; |
316 | ++enc->indent; |
302 | |
317 | |
303 | for (i = 0; i <= len; ++i) |
318 | for (i = 0; i <= len; ++i) |
304 | { |
319 | { |
|
|
320 | SV **svp = av_fetch (av, i, 0); |
|
|
321 | |
305 | encode_indent (enc); |
322 | encode_indent (enc); |
306 | encode_sv (enc, *av_fetch (av, i, 0)); |
323 | |
|
|
324 | if (svp) |
|
|
325 | encode_sv (enc, *svp); |
|
|
326 | else |
|
|
327 | encode_str (enc, "null", 4, 0); |
307 | |
328 | |
308 | if (i < len) |
329 | if (i < len) |
309 | encode_comma (enc); |
330 | encode_comma (enc); |
310 | } |
331 | } |
311 | |
332 | |
… | |
… | |
334 | else |
355 | else |
335 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
356 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
336 | |
357 | |
337 | encode_ch (enc, '"'); |
358 | encode_ch (enc, '"'); |
338 | |
359 | |
339 | if (enc->flags & F_SPACE_BEFORE) encode_space (enc); |
360 | if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); |
340 | encode_ch (enc, ':'); |
361 | encode_ch (enc, ':'); |
341 | if (enc->flags & F_SPACE_AFTER ) encode_space (enc); |
362 | if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); |
342 | encode_sv (enc, HeVAL (he)); |
363 | encode_sv (enc, HeVAL (he)); |
343 | } |
364 | } |
344 | |
365 | |
345 | // compare hash entries, used when all keys are bytestrings |
366 | // compare hash entries, used when all keys are bytestrings |
346 | static int |
367 | static int |
… | |
… | |
381 | { |
402 | { |
382 | // for canonical output we have to sort by keys first |
403 | // for canonical output we have to sort by keys first |
383 | // actually, this is mostly due to the stupid so-called |
404 | // actually, this is mostly due to the stupid so-called |
384 | // security workaround added somewhere in 5.8.x. |
405 | // security workaround added somewhere in 5.8.x. |
385 | // that randomises hash orderings |
406 | // that randomises hash orderings |
386 | if (enc->flags & F_CANONICAL) |
407 | if (enc->json.flags & F_CANONICAL) |
387 | { |
408 | { |
388 | int fast = 1; |
409 | int fast = 1; |
389 | HE *he; |
410 | HE *he; |
390 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
411 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
391 | HE **hes = _alloca (count * sizeof (HE)); |
412 | HE **hes = _alloca (count * sizeof (HE)); |
… | |
… | |
463 | svtype svt; |
484 | svtype svt; |
464 | |
485 | |
465 | SvGETMAGIC (sv); |
486 | SvGETMAGIC (sv); |
466 | svt = SvTYPE (sv); |
487 | svt = SvTYPE (sv); |
467 | |
488 | |
|
|
489 | if (expect_false (SvOBJECT (sv))) |
|
|
490 | { |
|
|
491 | HV *stash = !JSON_SLOW || json_boolean_stash |
|
|
492 | ? json_boolean_stash |
|
|
493 | : gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
494 | |
|
|
495 | if (SvSTASH (sv) == stash) |
|
|
496 | { |
|
|
497 | if (SvIV (sv)) |
|
|
498 | encode_str (enc, "true", 4, 0); |
|
|
499 | else |
|
|
500 | encode_str (enc, "false", 5, 0); |
|
|
501 | } |
|
|
502 | else |
|
|
503 | { |
|
|
504 | #if 0 |
|
|
505 | if (0 && sv_derived_from (rv, "JSON::Literal")) |
|
|
506 | { |
|
|
507 | // not yet |
|
|
508 | } |
|
|
509 | #endif |
|
|
510 | if (enc->json.flags & F_CONV_BLESSED) |
|
|
511 | { |
|
|
512 | // we re-bless the reference to get overload and other niceties right |
|
|
513 | GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0); |
|
|
514 | |
|
|
515 | if (to_json) |
|
|
516 | { |
|
|
517 | dSP; |
|
|
518 | |
|
|
519 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
520 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
|
|
521 | |
|
|
522 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
523 | PUTBACK; |
|
|
524 | call_sv ((SV *)GvCV (to_json), G_SCALAR); |
|
|
525 | SPAGAIN; |
|
|
526 | |
|
|
527 | // catch this surprisingly common error |
|
|
528 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
529 | croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
530 | |
|
|
531 | sv = POPs; |
|
|
532 | PUTBACK; |
|
|
533 | |
|
|
534 | encode_sv (enc, sv); |
|
|
535 | |
|
|
536 | FREETMPS; LEAVE; |
|
|
537 | } |
|
|
538 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
539 | encode_str (enc, "null", 4, 0); |
|
|
540 | else |
|
|
541 | croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", |
|
|
542 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
543 | } |
|
|
544 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
545 | encode_str (enc, "null", 4, 0); |
|
|
546 | else |
|
|
547 | croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", |
|
|
548 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
549 | } |
|
|
550 | } |
468 | if (svt == SVt_PVHV) |
551 | else if (svt == SVt_PVHV) |
469 | encode_hv (enc, (HV *)sv); |
552 | encode_hv (enc, (HV *)sv); |
470 | else if (svt == SVt_PVAV) |
553 | else if (svt == SVt_PVAV) |
471 | encode_av (enc, (AV *)sv); |
554 | encode_av (enc, (AV *)sv); |
472 | else if (svt < SVt_PVAV) |
555 | else if (svt < SVt_PVAV) |
473 | { |
556 | { |
474 | if (SvNIOK (sv) && SvIV (sv) == 0) |
557 | STRLEN len = 0; |
|
|
558 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
559 | |
|
|
560 | if (len == 1 && *pv == '1') |
|
|
561 | encode_str (enc, "true", 4, 0); |
|
|
562 | else if (len == 1 && *pv == '0') |
475 | encode_str (enc, "false", 5, 0); |
563 | encode_str (enc, "false", 5, 0); |
476 | else if (SvNIOK (sv) && SvIV (sv) == 1) |
|
|
477 | encode_str (enc, "true", 4, 0); |
|
|
478 | else |
564 | else |
479 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
565 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
480 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
566 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
481 | } |
567 | } |
482 | else |
568 | else |
… | |
… | |
497 | encode_str (enc, str, len, SvUTF8 (sv)); |
583 | encode_str (enc, str, len, SvUTF8 (sv)); |
498 | encode_ch (enc, '"'); |
584 | encode_ch (enc, '"'); |
499 | } |
585 | } |
500 | else if (SvNOKp (sv)) |
586 | else if (SvNOKp (sv)) |
501 | { |
587 | { |
|
|
588 | // trust that perl will do the right thing w.r.t. JSON syntax. |
502 | need (enc, NV_DIG + 32); |
589 | need (enc, NV_DIG + 32); |
503 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
590 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
504 | enc->cur += strlen (enc->cur); |
591 | enc->cur += strlen (enc->cur); |
505 | } |
592 | } |
506 | else if (SvIOKp (sv)) |
593 | else if (SvIOKp (sv)) |
507 | { |
594 | { |
508 | // we assume we can always read an IV as a UV |
595 | // we assume we can always read an IV as a UV |
509 | if (SvUV (sv) & ~(UV)0x7fff) |
596 | if (SvUV (sv) & ~(UV)0x7fff) |
510 | { |
597 | { |
|
|
598 | // large integer, use the (rather slow) snprintf way. |
511 | need (enc, sizeof (UV) * 3); |
599 | need (enc, sizeof (UV) * 3); |
512 | enc->cur += |
600 | enc->cur += |
513 | SvIsUV(sv) |
601 | SvIsUV(sv) |
514 | ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) |
602 | ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) |
515 | : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv)); |
603 | : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv)); |
… | |
… | |
518 | { |
606 | { |
519 | // optimise the "small number case" |
607 | // optimise the "small number case" |
520 | // code will likely be branchless and use only a single multiplication |
608 | // code will likely be branchless and use only a single multiplication |
521 | I32 i = SvIV (sv); |
609 | I32 i = SvIV (sv); |
522 | U32 u; |
610 | U32 u; |
|
|
611 | char digit, nz = 0; |
523 | |
612 | |
524 | need (enc, 6); |
613 | need (enc, 6); |
525 | |
614 | |
526 | *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; |
615 | *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; |
527 | u = i < 0 ? -i : i; |
616 | u = i < 0 ? -i : i; |
528 | |
617 | |
529 | // convert to 4.28 fixed-point representation |
618 | // convert to 4.28 fixed-point representation |
530 | u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits |
619 | u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits |
531 | |
620 | |
532 | char digit, nz = 0; |
621 | // now output digit by digit, each time masking out the integer part |
533 | |
622 | // and multiplying by 5 while moving the decimal point one to the right, |
|
|
623 | // resulting in a net multiplication by 10. |
|
|
624 | // we always write the digit to memory but conditionally increment |
|
|
625 | // the pointer, to ease the usage of conditional move instructions. |
534 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; |
626 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; |
535 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; |
627 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; |
536 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; |
628 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; |
537 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; |
629 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; |
538 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; |
630 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
539 | } |
631 | } |
540 | } |
632 | } |
541 | else if (SvROK (sv)) |
633 | else if (SvROK (sv)) |
542 | encode_rv (enc, SvRV (sv)); |
634 | encode_rv (enc, SvRV (sv)); |
543 | else if (!SvOK (sv)) |
635 | else if (!SvOK (sv)) |
… | |
… | |
546 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
638 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
547 | SvPV_nolen (sv), SvFLAGS (sv)); |
639 | SvPV_nolen (sv), SvFLAGS (sv)); |
548 | } |
640 | } |
549 | |
641 | |
550 | static SV * |
642 | static SV * |
551 | encode_json (SV *scalar, U32 flags) |
643 | encode_json (SV *scalar, JSON *json) |
552 | { |
644 | { |
553 | enc_t enc; |
645 | enc_t enc; |
554 | |
646 | |
555 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
647 | if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
556 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
648 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
557 | |
649 | |
558 | enc.flags = flags; |
650 | enc.json = *json; |
559 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
651 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
560 | enc.cur = SvPVX (enc.sv); |
652 | enc.cur = SvPVX (enc.sv); |
561 | enc.end = SvEND (enc.sv); |
653 | enc.end = SvEND (enc.sv); |
562 | enc.indent = 0; |
654 | enc.indent = 0; |
563 | enc.maxdepth = DEC_DEPTH (flags); |
655 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
564 | |
656 | |
565 | SvPOK_only (enc.sv); |
657 | SvPOK_only (enc.sv); |
566 | encode_sv (&enc, scalar); |
658 | encode_sv (&enc, scalar); |
567 | |
659 | |
568 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
660 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
569 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
661 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
570 | |
662 | |
571 | if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
663 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
572 | SvUTF8_on (enc.sv); |
664 | SvUTF8_on (enc.sv); |
573 | |
665 | |
574 | if (enc.flags & F_SHRINK) |
666 | if (enc.json.flags & F_SHRINK) |
575 | shrink (enc.sv); |
667 | shrink (enc.sv); |
576 | |
668 | |
577 | return enc.sv; |
669 | return enc.sv; |
578 | } |
670 | } |
579 | |
671 | |
… | |
… | |
584 | typedef struct |
676 | typedef struct |
585 | { |
677 | { |
586 | char *cur; // current parser pointer |
678 | char *cur; // current parser pointer |
587 | char *end; // end of input string |
679 | char *end; // end of input string |
588 | const char *err; // parse error, if != 0 |
680 | const char *err; // parse error, if != 0 |
589 | U32 flags; // F_* |
681 | JSON json; |
590 | U32 depth; // recursion depth |
682 | U32 depth; // recursion depth |
591 | U32 maxdepth; // recursion depth limit |
683 | U32 maxdepth; // recursion depth limit |
592 | } dec_t; |
684 | } dec_t; |
593 | |
685 | |
594 | inline void |
686 | inline void |
… | |
… | |
625 | decode_4hex (dec_t *dec) |
717 | decode_4hex (dec_t *dec) |
626 | { |
718 | { |
627 | signed char d1, d2, d3, d4; |
719 | signed char d1, d2, d3, d4; |
628 | unsigned char *cur = (unsigned char *)dec->cur; |
720 | unsigned char *cur = (unsigned char *)dec->cur; |
629 | |
721 | |
630 | d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected"); |
722 | d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected"); |
631 | d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected"); |
723 | d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected"); |
632 | d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected"); |
724 | d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected"); |
633 | d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected"); |
725 | d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected"); |
634 | |
726 | |
635 | dec->cur += 4; |
727 | dec->cur += 4; |
636 | |
728 | |
637 | return ((UV)d1) << 12 |
729 | return ((UV)d1) << 12 |
638 | | ((UV)d2) << 8 |
730 | | ((UV)d2) << 8 |
… | |
… | |
857 | is_nv = 1; |
949 | is_nv = 1; |
858 | } |
950 | } |
859 | |
951 | |
860 | if (!is_nv) |
952 | if (!is_nv) |
861 | { |
953 | { |
|
|
954 | int len = dec->cur - start; |
|
|
955 | |
862 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
956 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
863 | if (*start == '-') |
957 | if (*start == '-') |
864 | switch (dec->cur - start) |
958 | switch (len) |
865 | { |
959 | { |
866 | case 2: return newSViv (-( start [1] - '0' )); |
960 | case 2: return newSViv (-( start [1] - '0' * 1)); |
867 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
961 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
868 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
962 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
869 | case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
963 | case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
870 | } |
964 | } |
871 | else |
965 | else |
872 | switch (dec->cur - start) |
966 | switch (len) |
873 | { |
967 | { |
874 | case 1: return newSViv ( start [0] - '0' ); |
968 | case 1: return newSViv ( start [0] - '0' * 1); |
875 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
969 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
876 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
970 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
877 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
971 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
878 | } |
972 | } |
879 | |
973 | |
880 | { |
974 | { |
881 | UV uv; |
975 | UV uv; |
882 | int numtype = grok_number (start, dec->cur - start, &uv); |
976 | int numtype = grok_number (start, len, &uv); |
883 | if (numtype & IS_NUMBER_IN_UV) |
977 | if (numtype & IS_NUMBER_IN_UV) |
884 | if (numtype & IS_NUMBER_NEG) |
978 | if (numtype & IS_NUMBER_NEG) |
885 | { |
979 | { |
886 | if (uv < (UV)IV_MIN) |
980 | if (uv < (UV)IV_MIN) |
887 | return newSViv (-(IV)uv); |
981 | return newSViv (-(IV)uv); |
888 | } |
982 | } |
889 | else |
983 | else |
890 | return newSVuv (uv); |
984 | return newSVuv (uv); |
891 | } |
985 | } |
892 | } |
|
|
893 | |
986 | |
|
|
987 | len -= *start == '-' ? 1 : 0; |
|
|
988 | |
|
|
989 | // does not fit into IV or UV, try NV |
|
|
990 | if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) |
|
|
991 | #if defined (LDBL_DIG) |
|
|
992 | || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len) |
|
|
993 | #endif |
|
|
994 | ) |
|
|
995 | // fits into NV without loss of precision |
|
|
996 | return newSVnv (Atof (start)); |
|
|
997 | |
|
|
998 | // everything else fails, convert it to a string |
|
|
999 | return newSVpvn (start, dec->cur - start); |
|
|
1000 | } |
|
|
1001 | |
|
|
1002 | // loss of precision here |
894 | return newSVnv (Atof (start)); |
1003 | return newSVnv (Atof (start)); |
895 | |
1004 | |
896 | fail: |
1005 | fail: |
897 | return 0; |
1006 | return 0; |
898 | } |
1007 | } |
… | |
… | |
942 | } |
1051 | } |
943 | |
1052 | |
944 | static SV * |
1053 | static SV * |
945 | decode_hv (dec_t *dec) |
1054 | decode_hv (dec_t *dec) |
946 | { |
1055 | { |
|
|
1056 | SV *sv; |
947 | HV *hv = newHV (); |
1057 | HV *hv = newHV (); |
948 | |
1058 | |
949 | DEC_INC_DEPTH; |
1059 | DEC_INC_DEPTH; |
950 | decode_ws (dec); |
1060 | decode_ws (dec); |
951 | |
1061 | |
952 | if (*dec->cur == '}') |
1062 | if (*dec->cur == '}') |
953 | ++dec->cur; |
1063 | ++dec->cur; |
954 | else |
1064 | else |
955 | for (;;) |
1065 | for (;;) |
956 | { |
1066 | { |
957 | SV *key, *value; |
|
|
958 | |
|
|
959 | decode_ws (dec); EXPECT_CH ('"'); |
1067 | decode_ws (dec); EXPECT_CH ('"'); |
960 | |
1068 | |
961 | key = decode_str (dec); |
1069 | // heuristic: assume that |
962 | if (!key) |
1070 | // a) decode_str + hv_store_ent are abysmally slow. |
963 | goto fail; |
1071 | // b) most hash keys are short, simple ascii text. |
|
|
1072 | // => try to "fast-match" such strings to avoid |
|
|
1073 | // the overhead of decode_str + hv_store_ent. |
|
|
1074 | { |
|
|
1075 | SV *value; |
|
|
1076 | char *p = dec->cur; |
|
|
1077 | char *e = p + 24; // only try up to 24 bytes |
964 | |
1078 | |
965 | decode_ws (dec); EXPECT_CH (':'); |
1079 | for (;;) |
966 | |
|
|
967 | value = decode_sv (dec); |
|
|
968 | if (!value) |
|
|
969 | { |
1080 | { |
|
|
1081 | // the >= 0x80 is true on most architectures |
|
|
1082 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
|
|
1083 | { |
|
|
1084 | // slow path, back up and use decode_str |
|
|
1085 | SV *key = decode_str (dec); |
|
|
1086 | if (!key) |
|
|
1087 | goto fail; |
|
|
1088 | |
|
|
1089 | decode_ws (dec); EXPECT_CH (':'); |
|
|
1090 | |
|
|
1091 | value = decode_sv (dec); |
|
|
1092 | if (!value) |
|
|
1093 | { |
|
|
1094 | SvREFCNT_dec (key); |
|
|
1095 | goto fail; |
|
|
1096 | } |
|
|
1097 | |
|
|
1098 | hv_store_ent (hv, key, value, 0); |
970 | SvREFCNT_dec (key); |
1099 | SvREFCNT_dec (key); |
|
|
1100 | |
|
|
1101 | break; |
|
|
1102 | } |
|
|
1103 | else if (*p == '"') |
|
|
1104 | { |
|
|
1105 | // fast path, got a simple key |
|
|
1106 | char *key = dec->cur; |
|
|
1107 | int len = p - key; |
|
|
1108 | dec->cur = p + 1; |
|
|
1109 | |
|
|
1110 | decode_ws (dec); EXPECT_CH (':'); |
|
|
1111 | |
|
|
1112 | value = decode_sv (dec); |
|
|
1113 | if (!value) |
971 | goto fail; |
1114 | goto fail; |
|
|
1115 | |
|
|
1116 | hv_store (hv, key, len, value, 0); |
|
|
1117 | |
|
|
1118 | break; |
|
|
1119 | } |
|
|
1120 | |
|
|
1121 | ++p; |
972 | } |
1122 | } |
973 | |
1123 | } |
974 | hv_store_ent (hv, key, value, 0); |
|
|
975 | SvREFCNT_dec (key); |
|
|
976 | |
1124 | |
977 | decode_ws (dec); |
1125 | decode_ws (dec); |
978 | |
1126 | |
979 | if (*dec->cur == '}') |
1127 | if (*dec->cur == '}') |
980 | { |
1128 | { |
… | |
… | |
987 | |
1135 | |
988 | ++dec->cur; |
1136 | ++dec->cur; |
989 | } |
1137 | } |
990 | |
1138 | |
991 | DEC_DEC_DEPTH; |
1139 | DEC_DEC_DEPTH; |
992 | return newRV_noinc ((SV *)hv); |
1140 | sv = newRV_noinc ((SV *)hv); |
|
|
1141 | |
|
|
1142 | // check filter callbacks |
|
|
1143 | if (dec->json.flags & F_HOOK) |
|
|
1144 | { |
|
|
1145 | if (dec->json.cb_sk_object && HvKEYS (hv) == 1) |
|
|
1146 | { |
|
|
1147 | HE *cb, *he; |
|
|
1148 | |
|
|
1149 | hv_iterinit (hv); |
|
|
1150 | he = hv_iternext (hv); |
|
|
1151 | hv_iterinit (hv); |
|
|
1152 | |
|
|
1153 | // the next line creates a mortal sv each time its called. |
|
|
1154 | // might want to optimise this for common cases. |
|
|
1155 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
|
|
1156 | |
|
|
1157 | if (cb) |
|
|
1158 | { |
|
|
1159 | dSP; |
|
|
1160 | int count; |
|
|
1161 | |
|
|
1162 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
1163 | XPUSHs (HeVAL (he)); |
|
|
1164 | |
|
|
1165 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
|
|
1166 | |
|
|
1167 | if (count == 1) |
|
|
1168 | { |
|
|
1169 | sv = newSVsv (POPs); |
|
|
1170 | FREETMPS; LEAVE; |
|
|
1171 | return sv; |
|
|
1172 | } |
|
|
1173 | |
|
|
1174 | FREETMPS; LEAVE; |
|
|
1175 | } |
|
|
1176 | } |
|
|
1177 | |
|
|
1178 | if (dec->json.cb_object) |
|
|
1179 | { |
|
|
1180 | dSP; |
|
|
1181 | int count; |
|
|
1182 | |
|
|
1183 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
1184 | XPUSHs (sv_2mortal (sv)); |
|
|
1185 | |
|
|
1186 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
|
|
1187 | |
|
|
1188 | if (count == 1) |
|
|
1189 | { |
|
|
1190 | sv = newSVsv (POPs); |
|
|
1191 | FREETMPS; LEAVE; |
|
|
1192 | return sv; |
|
|
1193 | } |
|
|
1194 | |
|
|
1195 | SvREFCNT_inc (sv); |
|
|
1196 | FREETMPS; LEAVE; |
|
|
1197 | } |
|
|
1198 | } |
|
|
1199 | |
|
|
1200 | return sv; |
993 | |
1201 | |
994 | fail: |
1202 | fail: |
995 | SvREFCNT_dec (hv); |
1203 | SvREFCNT_dec (hv); |
996 | DEC_DEC_DEPTH; |
1204 | DEC_DEC_DEPTH; |
997 | return 0; |
1205 | return 0; |
… | |
… | |
999 | |
1207 | |
1000 | static SV * |
1208 | static SV * |
1001 | decode_sv (dec_t *dec) |
1209 | decode_sv (dec_t *dec) |
1002 | { |
1210 | { |
1003 | decode_ws (dec); |
1211 | decode_ws (dec); |
|
|
1212 | |
|
|
1213 | // the beauty of JSON: you need exactly one character lookahead |
|
|
1214 | // to parse anything. |
1004 | switch (*dec->cur) |
1215 | switch (*dec->cur) |
1005 | { |
1216 | { |
1006 | case '"': ++dec->cur; return decode_str (dec); |
1217 | case '"': ++dec->cur; return decode_str (dec); |
1007 | case '[': ++dec->cur; return decode_av (dec); |
1218 | case '[': ++dec->cur; return decode_av (dec); |
1008 | case '{': ++dec->cur; return decode_hv (dec); |
1219 | case '{': ++dec->cur; return decode_hv (dec); |
… | |
… | |
1014 | |
1225 | |
1015 | case 't': |
1226 | case 't': |
1016 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1227 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1017 | { |
1228 | { |
1018 | dec->cur += 4; |
1229 | dec->cur += 4; |
1019 | return newSViv (1); |
1230 | return SvREFCNT_inc (json_true); |
1020 | } |
1231 | } |
1021 | else |
1232 | else |
1022 | ERR ("'true' expected"); |
1233 | ERR ("'true' expected"); |
1023 | |
1234 | |
1024 | break; |
1235 | break; |
1025 | |
1236 | |
1026 | case 'f': |
1237 | case 'f': |
1027 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1238 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1028 | { |
1239 | { |
1029 | dec->cur += 5; |
1240 | dec->cur += 5; |
1030 | return newSViv (0); |
1241 | return SvREFCNT_inc (json_false); |
1031 | } |
1242 | } |
1032 | else |
1243 | else |
1033 | ERR ("'false' expected"); |
1244 | ERR ("'false' expected"); |
1034 | |
1245 | |
1035 | break; |
1246 | break; |
… | |
… | |
1053 | fail: |
1264 | fail: |
1054 | return 0; |
1265 | return 0; |
1055 | } |
1266 | } |
1056 | |
1267 | |
1057 | static SV * |
1268 | static SV * |
1058 | decode_json (SV *string, U32 flags, UV *offset_return) |
1269 | decode_json (SV *string, JSON *json, UV *offset_return) |
1059 | { |
1270 | { |
1060 | dec_t dec; |
1271 | dec_t dec; |
1061 | UV offset; |
1272 | UV offset; |
1062 | SV *sv; |
1273 | SV *sv; |
1063 | |
1274 | |
1064 | SvGETMAGIC (string); |
1275 | SvGETMAGIC (string); |
1065 | SvUPGRADE (string, SVt_PV); |
1276 | SvUPGRADE (string, SVt_PV); |
1066 | |
1277 | |
|
|
1278 | if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) |
|
|
1279 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
1280 | (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); |
|
|
1281 | |
1067 | if (flags & F_UTF8) |
1282 | if (json->flags & F_UTF8) |
1068 | sv_utf8_downgrade (string, 0); |
1283 | sv_utf8_downgrade (string, 0); |
1069 | else |
1284 | else |
1070 | sv_utf8_upgrade (string); |
1285 | sv_utf8_upgrade (string); |
1071 | |
1286 | |
1072 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1287 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1073 | |
1288 | |
1074 | dec.flags = flags; |
1289 | dec.json = *json; |
1075 | dec.cur = SvPVX (string); |
1290 | dec.cur = SvPVX (string); |
1076 | dec.end = SvEND (string); |
1291 | dec.end = SvEND (string); |
1077 | dec.err = 0; |
1292 | dec.err = 0; |
1078 | dec.depth = 0; |
1293 | dec.depth = 0; |
1079 | dec.maxdepth = DEC_DEPTH (dec.flags); |
1294 | dec.maxdepth = DEC_DEPTH (dec.json.flags); |
|
|
1295 | |
|
|
1296 | if (dec.json.cb_object || dec.json.cb_sk_object) |
|
|
1297 | dec.json.flags |= F_HOOK; |
1080 | |
1298 | |
1081 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1299 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1082 | sv = decode_sv (&dec); |
1300 | sv = decode_sv (&dec); |
1083 | |
1301 | |
1084 | if (!(offset_return || !sv)) |
1302 | if (!(offset_return || !sv)) |
… | |
… | |
1094 | } |
1312 | } |
1095 | } |
1313 | } |
1096 | |
1314 | |
1097 | if (offset_return || !sv) |
1315 | if (offset_return || !sv) |
1098 | { |
1316 | { |
1099 | offset = dec.flags & F_UTF8 |
1317 | offset = dec.json.flags & F_UTF8 |
1100 | ? dec.cur - SvPVX (string) |
1318 | ? dec.cur - SvPVX (string) |
1101 | : utf8_distance (dec.cur, SvPVX (string)); |
1319 | : utf8_distance (dec.cur, SvPVX (string)); |
1102 | |
1320 | |
1103 | if (offset_return) |
1321 | if (offset_return) |
1104 | *offset_return = offset; |
1322 | *offset_return = offset; |
… | |
… | |
1123 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1341 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1124 | } |
1342 | } |
1125 | |
1343 | |
1126 | sv = sv_2mortal (sv); |
1344 | sv = sv_2mortal (sv); |
1127 | |
1345 | |
1128 | if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1346 | if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1129 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1347 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1130 | |
1348 | |
1131 | return sv; |
1349 | return sv; |
1132 | } |
1350 | } |
1133 | |
1351 | |
… | |
… | |
1137 | MODULE = JSON::XS PACKAGE = JSON::XS |
1355 | MODULE = JSON::XS PACKAGE = JSON::XS |
1138 | |
1356 | |
1139 | BOOT: |
1357 | BOOT: |
1140 | { |
1358 | { |
1141 | int i; |
1359 | int i; |
1142 | |
|
|
1143 | memset (decode_hexdigit, 0xff, 256); |
|
|
1144 | |
1360 | |
1145 | for (i = 0; i < 256; ++i) |
1361 | for (i = 0; i < 256; ++i) |
1146 | decode_hexdigit [i] = |
1362 | decode_hexdigit [i] = |
1147 | i >= '0' && i <= '9' ? i - '0' |
1363 | i >= '0' && i <= '9' ? i - '0' |
1148 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1364 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1149 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1365 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1150 | : -1; |
1366 | : -1; |
1151 | |
1367 | |
1152 | json_stash = gv_stashpv ("JSON::XS", 1); |
1368 | json_stash = gv_stashpv ("JSON::XS" , 1); |
|
|
1369 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
1370 | |
|
|
1371 | json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
|
|
1372 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1153 | } |
1373 | } |
1154 | |
1374 | |
1155 | PROTOTYPES: DISABLE |
1375 | PROTOTYPES: DISABLE |
1156 | |
1376 | |
1157 | SV *new (char *dummy) |
1377 | void CLONE (...) |
1158 | CODE: |
1378 | CODE: |
1159 | RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); |
1379 | json_stash = 0; |
1160 | OUTPUT: |
1380 | json_boolean_stash = 0; |
1161 | RETVAL |
|
|
1162 | |
1381 | |
|
|
1382 | void new (char *klass) |
|
|
1383 | PPCODE: |
|
|
1384 | { |
|
|
1385 | SV *pv = NEWSV (0, sizeof (JSON)); |
|
|
1386 | SvPOK_only (pv); |
|
|
1387 | Zero (SvPVX (pv), 1, JSON); |
|
|
1388 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
|
|
1389 | XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH))); |
|
|
1390 | } |
|
|
1391 | |
1163 | SV *ascii (SV *self, int enable = 1) |
1392 | void ascii (JSON *self, int enable = 1) |
1164 | ALIAS: |
1393 | ALIAS: |
1165 | ascii = F_ASCII |
1394 | ascii = F_ASCII |
1166 | latin1 = F_LATIN1 |
1395 | latin1 = F_LATIN1 |
1167 | utf8 = F_UTF8 |
1396 | utf8 = F_UTF8 |
1168 | indent = F_INDENT |
1397 | indent = F_INDENT |
1169 | canonical = F_CANONICAL |
1398 | canonical = F_CANONICAL |
1170 | space_before = F_SPACE_BEFORE |
1399 | space_before = F_SPACE_BEFORE |
1171 | space_after = F_SPACE_AFTER |
1400 | space_after = F_SPACE_AFTER |
1172 | pretty = F_PRETTY |
1401 | pretty = F_PRETTY |
1173 | allow_nonref = F_ALLOW_NONREF |
1402 | allow_nonref = F_ALLOW_NONREF |
1174 | shrink = F_SHRINK |
1403 | shrink = F_SHRINK |
|
|
1404 | allow_blessed = F_ALLOW_BLESSED |
|
|
1405 | convert_blessed = F_CONV_BLESSED |
1175 | CODE: |
1406 | PPCODE: |
1176 | { |
1407 | { |
1177 | UV *uv = SvJSON (self); |
|
|
1178 | if (enable) |
1408 | if (enable) |
1179 | *uv |= ix; |
1409 | self->flags |= ix; |
1180 | else |
1410 | else |
1181 | *uv &= ~ix; |
1411 | self->flags &= ~ix; |
1182 | |
1412 | |
1183 | RETVAL = newSVsv (self); |
1413 | XPUSHs (ST (0)); |
1184 | } |
1414 | } |
1185 | OUTPUT: |
|
|
1186 | RETVAL |
|
|
1187 | |
1415 | |
1188 | SV *max_depth (SV *self, UV max_depth = 0x80000000UL) |
1416 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
1189 | CODE: |
1417 | PPCODE: |
1190 | { |
1418 | { |
1191 | UV *uv = SvJSON (self); |
|
|
1192 | UV log2 = 0; |
1419 | UV log2 = 0; |
1193 | |
1420 | |
1194 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
1421 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
1195 | |
1422 | |
1196 | while ((1UL << log2) < max_depth) |
1423 | while ((1UL << log2) < max_depth) |
1197 | ++log2; |
1424 | ++log2; |
1198 | |
1425 | |
1199 | *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
1426 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
1200 | |
1427 | |
1201 | RETVAL = newSVsv (self); |
1428 | XPUSHs (ST (0)); |
1202 | } |
1429 | } |
1203 | OUTPUT: |
|
|
1204 | RETVAL |
|
|
1205 | |
1430 | |
1206 | void encode (SV *self, SV *scalar) |
1431 | void max_size (JSON *self, UV max_size = 0) |
1207 | PPCODE: |
1432 | PPCODE: |
1208 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1433 | { |
|
|
1434 | UV log2 = 0; |
1209 | |
1435 | |
1210 | void decode (SV *self, SV *jsonstr) |
1436 | if (max_size > 0x80000000UL) max_size = 0x80000000UL; |
|
|
1437 | if (max_size == 1) max_size = 2; |
|
|
1438 | |
|
|
1439 | while ((1UL << log2) < max_size) |
|
|
1440 | ++log2; |
|
|
1441 | |
|
|
1442 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
|
|
1443 | |
|
|
1444 | XPUSHs (ST (0)); |
|
|
1445 | } |
|
|
1446 | |
|
|
1447 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1211 | PPCODE: |
1448 | PPCODE: |
|
|
1449 | { |
|
|
1450 | SvREFCNT_dec (self->cb_object); |
|
|
1451 | self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; |
|
|
1452 | |
|
|
1453 | XPUSHs (ST (0)); |
|
|
1454 | } |
|
|
1455 | |
|
|
1456 | void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
|
|
1457 | PPCODE: |
|
|
1458 | { |
|
|
1459 | if (!self->cb_sk_object) |
|
|
1460 | self->cb_sk_object = newHV (); |
|
|
1461 | |
|
|
1462 | if (SvOK (cb)) |
|
|
1463 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
|
|
1464 | else |
|
|
1465 | { |
|
|
1466 | hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0); |
|
|
1467 | |
|
|
1468 | if (!HvKEYS (self->cb_sk_object)) |
|
|
1469 | { |
|
|
1470 | SvREFCNT_dec (self->cb_sk_object); |
|
|
1471 | self->cb_sk_object = 0; |
|
|
1472 | } |
|
|
1473 | } |
|
|
1474 | |
|
|
1475 | XPUSHs (ST (0)); |
|
|
1476 | } |
|
|
1477 | |
|
|
1478 | void encode (JSON *self, SV *scalar) |
|
|
1479 | PPCODE: |
|
|
1480 | XPUSHs (encode_json (scalar, self)); |
|
|
1481 | |
|
|
1482 | void decode (JSON *self, SV *jsonstr) |
|
|
1483 | PPCODE: |
1212 | XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); |
1484 | XPUSHs (decode_json (jsonstr, self, 0)); |
1213 | |
1485 | |
1214 | void decode_prefix (SV *self, SV *jsonstr) |
1486 | void decode_prefix (JSON *self, SV *jsonstr) |
1215 | PPCODE: |
1487 | PPCODE: |
1216 | { |
1488 | { |
1217 | UV offset; |
1489 | UV offset; |
1218 | EXTEND (SP, 2); |
1490 | EXTEND (SP, 2); |
1219 | PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); |
1491 | PUSHs (decode_json (jsonstr, self, &offset)); |
1220 | PUSHs (sv_2mortal (newSVuv (offset))); |
1492 | PUSHs (sv_2mortal (newSVuv (offset))); |
1221 | } |
1493 | } |
1222 | |
1494 | |
|
|
1495 | void DESTROY (JSON *self) |
|
|
1496 | CODE: |
|
|
1497 | SvREFCNT_dec (self->cb_sk_object); |
|
|
1498 | SvREFCNT_dec (self->cb_object); |
|
|
1499 | |
1223 | PROTOTYPES: ENABLE |
1500 | PROTOTYPES: ENABLE |
1224 | |
1501 | |
1225 | void to_json (SV *scalar) |
1502 | void to_json (SV *scalar) |
1226 | ALIAS: |
|
|
1227 | objToJson = 0 |
|
|
1228 | PPCODE: |
1503 | PPCODE: |
|
|
1504 | { |
|
|
1505 | JSON json = { F_DEFAULT | F_UTF8 }; |
1229 | XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); |
1506 | XPUSHs (encode_json (scalar, &json)); |
|
|
1507 | } |
1230 | |
1508 | |
1231 | void from_json (SV *jsonstr) |
1509 | void from_json (SV *jsonstr) |
1232 | ALIAS: |
|
|
1233 | jsonToObj = 0 |
|
|
1234 | PPCODE: |
1510 | PPCODE: |
|
|
1511 | { |
|
|
1512 | JSON json = { F_DEFAULT | F_UTF8 }; |
1235 | XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); |
1513 | XPUSHs (decode_json (jsonstr, &json, 0)); |
|
|
1514 | } |
1236 | |
1515 | |