1 |
#include "EXTERN.h" |
2 |
#include "perl.h" |
3 |
#include "XSUB.h" |
4 |
|
5 |
#include "assert.h" |
6 |
#include "string.h" |
7 |
#include "stdlib.h" |
8 |
|
9 |
#define F_ASCII 0x00000001 |
10 |
#define F_UTF8 0x00000002 |
11 |
#define F_INDENT 0x00000004 |
12 |
#define F_CANONICAL 0x00000008 |
13 |
#define F_SPACE_BEFORE 0x00000010 |
14 |
#define F_SPACE_AFTER 0x00000020 |
15 |
#define F_JSON_RPC 0x00000040 |
16 |
#define F_ALLOW_NONREF 0x00000080 |
17 |
#define F_SHRINK 0x00000100 |
18 |
|
19 |
#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
20 |
#define F_DEFAULT 0 |
21 |
|
22 |
#define INIT_SIZE 32 // initial scalar size to be allocated |
23 |
|
24 |
#define SB do { |
25 |
#define SE } while (0) |
26 |
|
27 |
static HV *json_stash; |
28 |
|
29 |
// structure used for encoding JSON |
30 |
typedef struct |
31 |
{ |
32 |
char *cur; |
33 |
STRLEN len; // SvLEN (sv) |
34 |
char *end; // SvEND (sv) |
35 |
SV *sv; |
36 |
UV flags; |
37 |
int max_recurse; |
38 |
int indent; |
39 |
} enc_t; |
40 |
|
41 |
// structure used for decoding JSON |
42 |
typedef struct |
43 |
{ |
44 |
char *cur; |
45 |
char *end; |
46 |
const char *err; |
47 |
UV flags; |
48 |
} dec_t; |
49 |
|
50 |
static UV * |
51 |
SvJSON (SV *sv) |
52 |
{ |
53 |
if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) |
54 |
croak ("object is not of type JSON::XS"); |
55 |
|
56 |
return &SvUVX (SvRV (sv)); |
57 |
} |
58 |
|
59 |
static void |
60 |
shrink (SV *sv) |
61 |
{ |
62 |
sv_utf8_downgrade (sv, 1); |
63 |
#ifdef SvPV_shrink_to_cur |
64 |
SvPV_shrink_to_cur (sv); |
65 |
#endif |
66 |
} |
67 |
|
68 |
///////////////////////////////////////////////////////////////////////////// |
69 |
|
70 |
static void |
71 |
need (enc_t *enc, STRLEN len) |
72 |
{ |
73 |
if (enc->cur + len >= enc->end) |
74 |
{ |
75 |
STRLEN cur = enc->cur - SvPVX (enc->sv); |
76 |
SvGROW (enc->sv, cur + len + 1); |
77 |
enc->cur = SvPVX (enc->sv) + cur; |
78 |
enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); |
79 |
} |
80 |
} |
81 |
|
82 |
static void |
83 |
encode_ch (enc_t *enc, char ch) |
84 |
{ |
85 |
need (enc, 1); |
86 |
*enc->cur++ = ch; |
87 |
} |
88 |
|
89 |
static void |
90 |
encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) |
91 |
{ |
92 |
char *end = str + len; |
93 |
|
94 |
need (enc, len); |
95 |
|
96 |
while (str < end) |
97 |
{ |
98 |
unsigned char ch = *(unsigned char *)str; |
99 |
|
100 |
if (ch >= 0x20 && ch < 0x80) // most common case |
101 |
{ |
102 |
if (ch == '"') // but with slow exceptions |
103 |
{ |
104 |
need (enc, len += 1); |
105 |
*enc->cur++ = '\\'; |
106 |
*enc->cur++ = '"'; |
107 |
} |
108 |
else if (ch == '\\') |
109 |
{ |
110 |
need (enc, len += 1); |
111 |
*enc->cur++ = '\\'; |
112 |
*enc->cur++ = '\\'; |
113 |
} |
114 |
else |
115 |
*enc->cur++ = ch; |
116 |
|
117 |
++str; |
118 |
} |
119 |
else |
120 |
{ |
121 |
switch (ch) |
122 |
{ |
123 |
case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
124 |
case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
125 |
case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
126 |
case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
127 |
case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
128 |
|
129 |
default: |
130 |
{ |
131 |
STRLEN clen; |
132 |
UV uch; |
133 |
|
134 |
if (is_utf8) |
135 |
{ |
136 |
uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
137 |
if (clen == (STRLEN)-1) |
138 |
croak ("malformed UTF-8 character in string, cannot convert to JSON"); |
139 |
} |
140 |
else |
141 |
{ |
142 |
uch = ch; |
143 |
clen = 1; |
144 |
} |
145 |
|
146 |
if (uch < 0x80 || enc->flags & F_ASCII) |
147 |
{ |
148 |
if (uch > 0xFFFFUL) |
149 |
{ |
150 |
need (enc, len += 11); |
151 |
sprintf (enc->cur, "\\u%04x\\u%04x", |
152 |
(uch - 0x10000) / 0x400 + 0xD800, |
153 |
(uch - 0x10000) % 0x400 + 0xDC00); |
154 |
enc->cur += 12; |
155 |
} |
156 |
else |
157 |
{ |
158 |
static char hexdigit [16] = "0123456789abcdef"; |
159 |
need (enc, len += 5); |
160 |
*enc->cur++ = '\\'; |
161 |
*enc->cur++ = 'u'; |
162 |
*enc->cur++ = hexdigit [ uch >> 12 ]; |
163 |
*enc->cur++ = hexdigit [(uch >> 8) & 15]; |
164 |
*enc->cur++ = hexdigit [(uch >> 4) & 15]; |
165 |
*enc->cur++ = hexdigit [(uch >> 0) & 15]; |
166 |
} |
167 |
|
168 |
str += clen; |
169 |
} |
170 |
else if (is_utf8) |
171 |
{ |
172 |
need (enc, len += clen); |
173 |
do |
174 |
{ |
175 |
*enc->cur++ = *str++; |
176 |
} |
177 |
while (--clen); |
178 |
} |
179 |
else |
180 |
{ |
181 |
need (enc, len += 10); // never more than 11 bytes needed |
182 |
enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
183 |
++str; |
184 |
} |
185 |
} |
186 |
} |
187 |
} |
188 |
|
189 |
--len; |
190 |
} |
191 |
} |
192 |
|
193 |
#define INDENT SB \ |
194 |
if (enc->flags & F_INDENT) \ |
195 |
{ \ |
196 |
int i_; \ |
197 |
need (enc, enc->indent); \ |
198 |
for (i_ = enc->indent * 3; i_--; )\ |
199 |
encode_ch (enc, ' '); \ |
200 |
} \ |
201 |
SE |
202 |
|
203 |
#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE |
204 |
#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE |
205 |
#define COMMA SB \ |
206 |
encode_ch (enc, ','); \ |
207 |
if (enc->flags & F_INDENT) \ |
208 |
NL; \ |
209 |
else if (enc->flags & F_SPACE_AFTER) \ |
210 |
SPACE; \ |
211 |
SE |
212 |
|
213 |
static void encode_sv (enc_t *enc, SV *sv); |
214 |
|
215 |
static void |
216 |
encode_av (enc_t *enc, AV *av) |
217 |
{ |
218 |
int i, len = av_len (av); |
219 |
|
220 |
encode_ch (enc, '['); NL; |
221 |
++enc->indent; |
222 |
|
223 |
for (i = 0; i <= len; ++i) |
224 |
{ |
225 |
INDENT; |
226 |
encode_sv (enc, *av_fetch (av, i, 0)); |
227 |
|
228 |
if (i < len) |
229 |
COMMA; |
230 |
} |
231 |
|
232 |
NL; |
233 |
|
234 |
--enc->indent; |
235 |
INDENT; encode_ch (enc, ']'); |
236 |
} |
237 |
|
238 |
static void |
239 |
encode_he (enc_t *enc, HE *he) |
240 |
{ |
241 |
encode_ch (enc, '"'); |
242 |
|
243 |
if (HeKLEN (he) == HEf_SVKEY) |
244 |
{ |
245 |
SV *sv = HeSVKEY (he); |
246 |
STRLEN len; |
247 |
char *str; |
248 |
|
249 |
SvGETMAGIC (sv); |
250 |
str = SvPV (sv, len); |
251 |
|
252 |
encode_str (enc, str, len, SvUTF8 (sv)); |
253 |
} |
254 |
else |
255 |
encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
256 |
|
257 |
encode_ch (enc, '"'); |
258 |
|
259 |
if (enc->flags & F_SPACE_BEFORE) SPACE; |
260 |
encode_ch (enc, ':'); |
261 |
if (enc->flags & F_SPACE_AFTER ) SPACE; |
262 |
encode_sv (enc, HeVAL (he)); |
263 |
} |
264 |
|
265 |
// compare hash entries, used when all keys are bytestrings |
266 |
static int |
267 |
he_cmp_fast (const void *a_, const void *b_) |
268 |
{ |
269 |
int cmp; |
270 |
|
271 |
HE *a = *(HE **)a_; |
272 |
HE *b = *(HE **)b_; |
273 |
|
274 |
STRLEN la = HeKLEN (a); |
275 |
STRLEN lb = HeKLEN (b); |
276 |
|
277 |
if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) |
278 |
cmp = la < lb ? -1 : la == lb ? 0 : 1; |
279 |
|
280 |
return cmp; |
281 |
} |
282 |
|
283 |
// compare hash entries, used when some keys are sv's or utf-x |
284 |
static int |
285 |
he_cmp_slow (const void *a, const void *b) |
286 |
{ |
287 |
return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); |
288 |
} |
289 |
|
290 |
static void |
291 |
encode_hv (enc_t *enc, HV *hv) |
292 |
{ |
293 |
int count, i; |
294 |
|
295 |
encode_ch (enc, '{'); NL; ++enc->indent; |
296 |
|
297 |
if ((count = hv_iterinit (hv))) |
298 |
{ |
299 |
// for canonical output we have to sort by keys first |
300 |
// actually, this is mostly due to the stupid so-called |
301 |
// security workaround added somewhere in 5.8.x. |
302 |
// that randomises hash orderings |
303 |
if (enc->flags & F_CANONICAL) |
304 |
{ |
305 |
HE *he, *hes [count]; |
306 |
int fast = 1; |
307 |
|
308 |
i = 0; |
309 |
while ((he = hv_iternext (hv))) |
310 |
{ |
311 |
hes [i++] = he; |
312 |
if (HeKLEN (he) < 0 || HeKUTF8 (he)) |
313 |
fast = 0; |
314 |
} |
315 |
|
316 |
assert (i == count); |
317 |
|
318 |
if (fast) |
319 |
qsort (hes, count, sizeof (HE *), he_cmp_fast); |
320 |
else |
321 |
{ |
322 |
// hack to disable "use bytes" |
323 |
COP *oldcop = PL_curcop, cop; |
324 |
cop.op_private = 0; |
325 |
PL_curcop = &cop; |
326 |
|
327 |
SAVETMPS; |
328 |
qsort (hes, count, sizeof (HE *), he_cmp_slow); |
329 |
FREETMPS; |
330 |
|
331 |
PL_curcop = oldcop; |
332 |
} |
333 |
|
334 |
for (i = 0; i < count; ++i) |
335 |
{ |
336 |
INDENT; |
337 |
encode_he (enc, hes [i]); |
338 |
|
339 |
if (i < count - 1) |
340 |
COMMA; |
341 |
} |
342 |
|
343 |
NL; |
344 |
} |
345 |
else |
346 |
{ |
347 |
SV *sv; |
348 |
HE *he = hv_iternext (hv); |
349 |
|
350 |
for (;;) |
351 |
{ |
352 |
INDENT; |
353 |
encode_he (enc, he); |
354 |
|
355 |
if (!(he = hv_iternext (hv))) |
356 |
break; |
357 |
|
358 |
COMMA; |
359 |
} |
360 |
|
361 |
NL; |
362 |
} |
363 |
} |
364 |
|
365 |
--enc->indent; INDENT; encode_ch (enc, '}'); |
366 |
} |
367 |
|
368 |
static void |
369 |
encode_sv (enc_t *enc, SV *sv) |
370 |
{ |
371 |
SvGETMAGIC (sv); |
372 |
|
373 |
if (SvPOKp (sv)) |
374 |
{ |
375 |
STRLEN len; |
376 |
char *str = SvPV (sv, len); |
377 |
encode_ch (enc, '"'); |
378 |
encode_str (enc, str, len, SvUTF8 (sv)); |
379 |
encode_ch (enc, '"'); |
380 |
} |
381 |
else if (SvNOKp (sv)) |
382 |
{ |
383 |
need (enc, NV_DIG + 32); |
384 |
Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
385 |
enc->cur += strlen (enc->cur); |
386 |
} |
387 |
else if (SvIOKp (sv)) |
388 |
{ |
389 |
need (enc, 64); |
390 |
enc->cur += |
391 |
SvIsUV(sv) |
392 |
? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) |
393 |
: snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); |
394 |
} |
395 |
else if (SvROK (sv)) |
396 |
{ |
397 |
if (!--enc->max_recurse) |
398 |
croak ("data structure too deep (hit recursion limit)"); |
399 |
|
400 |
sv = SvRV (sv); |
401 |
|
402 |
switch (SvTYPE (sv)) |
403 |
{ |
404 |
case SVt_PVAV: encode_av (enc, (AV *)sv); break; |
405 |
case SVt_PVHV: encode_hv (enc, (HV *)sv); break; |
406 |
|
407 |
default: |
408 |
croak ("JSON can only represent references to arrays or hashes"); |
409 |
} |
410 |
} |
411 |
else if (!SvOK (sv)) |
412 |
encode_str (enc, "null", 4, 0); |
413 |
else |
414 |
croak ("encountered perl type that JSON cannot handle"); |
415 |
} |
416 |
|
417 |
static SV * |
418 |
encode_json (SV *scalar, UV flags) |
419 |
{ |
420 |
if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
421 |
croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); |
422 |
|
423 |
enc_t enc; |
424 |
enc.flags = flags; |
425 |
enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
426 |
enc.cur = SvPVX (enc.sv); |
427 |
enc.end = SvEND (enc.sv); |
428 |
enc.max_recurse = 0; |
429 |
enc.indent = 0; |
430 |
|
431 |
SvPOK_only (enc.sv); |
432 |
encode_sv (&enc, scalar); |
433 |
|
434 |
if (!(flags & (F_ASCII | F_UTF8))) |
435 |
SvUTF8_on (enc.sv); |
436 |
|
437 |
SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
438 |
|
439 |
if (enc.flags & F_SHRINK) |
440 |
shrink (enc.sv); |
441 |
|
442 |
return enc.sv; |
443 |
} |
444 |
|
445 |
///////////////////////////////////////////////////////////////////////////// |
446 |
|
447 |
#define WS \ |
448 |
for (;;) \ |
449 |
{ \ |
450 |
char ch = *dec->cur; \ |
451 |
if (ch > 0x20 \ |
452 |
|| (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ |
453 |
break; \ |
454 |
++dec->cur; \ |
455 |
} |
456 |
|
457 |
#define ERR(reason) SB dec->err = reason; goto fail; SE |
458 |
#define EXPECT_CH(ch) SB \ |
459 |
if (*dec->cur != ch) \ |
460 |
ERR (# ch " expected"); \ |
461 |
++dec->cur; \ |
462 |
SE |
463 |
|
464 |
static SV *decode_sv (dec_t *dec); |
465 |
|
466 |
static signed char decode_hexdigit[256]; |
467 |
|
468 |
static UV |
469 |
decode_4hex (dec_t *dec) |
470 |
{ |
471 |
signed char d1, d2, d3, d4; |
472 |
|
473 |
d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; |
474 |
if (d1 < 0) ERR ("four hexadecimal digits expected"); |
475 |
d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; |
476 |
if (d2 < 0) ERR ("four hexadecimal digits expected"); |
477 |
d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]]; |
478 |
if (d3 < 0) ERR ("four hexadecimal digits expected"); |
479 |
d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]]; |
480 |
if (d4 < 0) ERR ("four hexadecimal digits expected"); |
481 |
|
482 |
dec->cur += 4; |
483 |
|
484 |
return ((UV)d1) << 12 |
485 |
| ((UV)d2) << 8 |
486 |
| ((UV)d3) << 4 |
487 |
| ((UV)d4); |
488 |
|
489 |
fail: |
490 |
return (UV)-1; |
491 |
} |
492 |
|
493 |
#define APPEND_GROW(n) SB \ |
494 |
if (cur + (n) >= end) \ |
495 |
{ \ |
496 |
STRLEN ofs = cur - SvPVX (sv); \ |
497 |
SvGROW (sv, ofs + (n) + 1); \ |
498 |
cur = SvPVX (sv) + ofs; \ |
499 |
end = SvEND (sv); \ |
500 |
} \ |
501 |
SE |
502 |
|
503 |
#define APPEND_CH(ch) SB \ |
504 |
APPEND_GROW (1); \ |
505 |
*cur++ = (ch); \ |
506 |
SE |
507 |
|
508 |
static SV * |
509 |
decode_str (dec_t *dec) |
510 |
{ |
511 |
SV *sv = NEWSV (0,2); |
512 |
int utf8 = 0; |
513 |
char *cur = SvPVX (sv); |
514 |
char *end = SvEND (sv); |
515 |
|
516 |
for (;;) |
517 |
{ |
518 |
unsigned char ch = *(unsigned char *)dec->cur; |
519 |
|
520 |
if (ch == '"') |
521 |
break; |
522 |
else if (ch == '\\') |
523 |
{ |
524 |
switch (*++dec->cur) |
525 |
{ |
526 |
case '\\': |
527 |
case '/': |
528 |
case '"': APPEND_CH (*dec->cur++); break; |
529 |
|
530 |
case 'b': APPEND_CH ('\010'); ++dec->cur; break; |
531 |
case 't': APPEND_CH ('\011'); ++dec->cur; break; |
532 |
case 'n': APPEND_CH ('\012'); ++dec->cur; break; |
533 |
case 'f': APPEND_CH ('\014'); ++dec->cur; break; |
534 |
case 'r': APPEND_CH ('\015'); ++dec->cur; break; |
535 |
|
536 |
case 'u': |
537 |
{ |
538 |
UV lo, hi; |
539 |
++dec->cur; |
540 |
|
541 |
hi = decode_4hex (dec); |
542 |
if (hi == (UV)-1) |
543 |
goto fail; |
544 |
|
545 |
// possibly a surrogate pair |
546 |
if (hi >= 0xd800 && hi < 0xdc00) |
547 |
{ |
548 |
if (dec->cur [0] != '\\' || dec->cur [1] != 'u') |
549 |
ERR ("missing low surrogate character in surrogate pair"); |
550 |
|
551 |
dec->cur += 2; |
552 |
|
553 |
lo = decode_4hex (dec); |
554 |
if (lo == (UV)-1) |
555 |
goto fail; |
556 |
|
557 |
if (lo < 0xdc00 || lo >= 0xe000) |
558 |
ERR ("surrogate pair expected"); |
559 |
|
560 |
hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; |
561 |
} |
562 |
else if (hi >= 0xdc00 && hi < 0xe000) |
563 |
ERR ("missing high surrogate character in surrogate pair"); |
564 |
|
565 |
if (hi >= 0x80) |
566 |
{ |
567 |
utf8 = 1; |
568 |
|
569 |
APPEND_GROW (4); // at most 4 bytes for 21 bits |
570 |
cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); |
571 |
} |
572 |
else |
573 |
APPEND_CH (hi); |
574 |
} |
575 |
break; |
576 |
|
577 |
default: |
578 |
--dec->cur; |
579 |
ERR ("illegal backslash escape sequence in string"); |
580 |
} |
581 |
} |
582 |
else if (ch >= 0x20 && ch <= 0x7f) |
583 |
APPEND_CH (*dec->cur++); |
584 |
else if (ch >= 0x80) |
585 |
{ |
586 |
STRLEN clen; |
587 |
UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); |
588 |
if (clen == (STRLEN)-1) |
589 |
ERR ("malformed UTF-8 character in JSON string"); |
590 |
|
591 |
APPEND_GROW (clen); |
592 |
do |
593 |
{ |
594 |
*cur++ = *dec->cur++; |
595 |
} |
596 |
while (--clen); |
597 |
|
598 |
utf8 = 1; |
599 |
} |
600 |
else if (dec->cur == dec->end) |
601 |
ERR ("unexpected end of string while parsing json string"); |
602 |
else |
603 |
ERR ("invalid character encountered"); |
604 |
} |
605 |
|
606 |
++dec->cur; |
607 |
|
608 |
SvCUR_set (sv, cur - SvPVX (sv)); |
609 |
|
610 |
SvPOK_only (sv); |
611 |
*SvEND (sv) = 0; |
612 |
|
613 |
if (utf8) |
614 |
SvUTF8_on (sv); |
615 |
|
616 |
if (dec->flags & F_SHRINK) |
617 |
shrink (sv); |
618 |
|
619 |
return sv; |
620 |
|
621 |
fail: |
622 |
SvREFCNT_dec (sv); |
623 |
return 0; |
624 |
} |
625 |
|
626 |
static SV * |
627 |
decode_num (dec_t *dec) |
628 |
{ |
629 |
int is_nv = 0; |
630 |
char *start = dec->cur; |
631 |
|
632 |
// [minus] |
633 |
if (*dec->cur == '-') |
634 |
++dec->cur; |
635 |
|
636 |
if (*dec->cur == '0') |
637 |
{ |
638 |
++dec->cur; |
639 |
if (*dec->cur >= '0' && *dec->cur <= '9') |
640 |
ERR ("malformed number (leading zero must not be followed by another digit)"); |
641 |
} |
642 |
else if (*dec->cur < '0' || *dec->cur > '9') |
643 |
ERR ("malformed number (no digits after initial minus)"); |
644 |
else |
645 |
do |
646 |
{ |
647 |
++dec->cur; |
648 |
} |
649 |
while (*dec->cur >= '0' && *dec->cur <= '9'); |
650 |
|
651 |
// [frac] |
652 |
if (*dec->cur == '.') |
653 |
{ |
654 |
++dec->cur; |
655 |
|
656 |
if (*dec->cur < '0' || *dec->cur > '9') |
657 |
ERR ("malformed number (no digits after decimal point)"); |
658 |
|
659 |
do |
660 |
{ |
661 |
++dec->cur; |
662 |
} |
663 |
while (*dec->cur >= '0' && *dec->cur <= '9'); |
664 |
|
665 |
is_nv = 1; |
666 |
} |
667 |
|
668 |
// [exp] |
669 |
if (*dec->cur == 'e' || *dec->cur == 'E') |
670 |
{ |
671 |
++dec->cur; |
672 |
|
673 |
if (*dec->cur == '-' || *dec->cur == '+') |
674 |
++dec->cur; |
675 |
|
676 |
if (*dec->cur < '0' || *dec->cur > '9') |
677 |
ERR ("malformed number (no digits after exp sign)"); |
678 |
|
679 |
do |
680 |
{ |
681 |
++dec->cur; |
682 |
} |
683 |
while (*dec->cur >= '0' && *dec->cur <= '9'); |
684 |
|
685 |
is_nv = 1; |
686 |
} |
687 |
|
688 |
if (!is_nv) |
689 |
{ |
690 |
UV uv; |
691 |
int numtype = grok_number (start, dec->cur - start, &uv); |
692 |
if (numtype & IS_NUMBER_IN_UV) |
693 |
if (numtype & IS_NUMBER_NEG) |
694 |
{ |
695 |
if (uv < (UV)IV_MIN) |
696 |
return newSViv (-(IV)uv); |
697 |
} |
698 |
else |
699 |
return newSVuv (uv); |
700 |
} |
701 |
|
702 |
return newSVnv (Atof (start)); |
703 |
|
704 |
fail: |
705 |
return 0; |
706 |
} |
707 |
|
708 |
static SV * |
709 |
decode_av (dec_t *dec) |
710 |
{ |
711 |
AV *av = newAV (); |
712 |
|
713 |
WS; |
714 |
if (*dec->cur == ']') |
715 |
++dec->cur; |
716 |
else |
717 |
for (;;) |
718 |
{ |
719 |
SV *value; |
720 |
|
721 |
value = decode_sv (dec); |
722 |
if (!value) |
723 |
goto fail; |
724 |
|
725 |
av_push (av, value); |
726 |
|
727 |
WS; |
728 |
|
729 |
if (*dec->cur == ']') |
730 |
{ |
731 |
++dec->cur; |
732 |
break; |
733 |
} |
734 |
|
735 |
if (*dec->cur != ',') |
736 |
ERR (", or ] expected while parsing array"); |
737 |
|
738 |
++dec->cur; |
739 |
} |
740 |
|
741 |
return newRV_noinc ((SV *)av); |
742 |
|
743 |
fail: |
744 |
SvREFCNT_dec (av); |
745 |
return 0; |
746 |
} |
747 |
|
748 |
static SV * |
749 |
decode_hv (dec_t *dec) |
750 |
{ |
751 |
HV *hv = newHV (); |
752 |
|
753 |
WS; |
754 |
if (*dec->cur == '}') |
755 |
++dec->cur; |
756 |
else |
757 |
for (;;) |
758 |
{ |
759 |
SV *key, *value; |
760 |
|
761 |
WS; EXPECT_CH ('"'); |
762 |
|
763 |
key = decode_str (dec); |
764 |
if (!key) |
765 |
goto fail; |
766 |
|
767 |
WS; EXPECT_CH (':'); |
768 |
|
769 |
value = decode_sv (dec); |
770 |
if (!value) |
771 |
{ |
772 |
SvREFCNT_dec (key); |
773 |
goto fail; |
774 |
} |
775 |
|
776 |
//TODO: optimise |
777 |
hv_store_ent (hv, key, value, 0); |
778 |
|
779 |
WS; |
780 |
|
781 |
if (*dec->cur == '}') |
782 |
{ |
783 |
++dec->cur; |
784 |
break; |
785 |
} |
786 |
|
787 |
if (*dec->cur != ',') |
788 |
ERR (", or } expected while parsing object/hash"); |
789 |
|
790 |
++dec->cur; |
791 |
} |
792 |
|
793 |
return newRV_noinc ((SV *)hv); |
794 |
|
795 |
fail: |
796 |
SvREFCNT_dec (hv); |
797 |
return 0; |
798 |
} |
799 |
|
800 |
static SV * |
801 |
decode_sv (dec_t *dec) |
802 |
{ |
803 |
WS; |
804 |
switch (*dec->cur) |
805 |
{ |
806 |
case '"': ++dec->cur; return decode_str (dec); |
807 |
case '[': ++dec->cur; return decode_av (dec); |
808 |
case '{': ++dec->cur; return decode_hv (dec); |
809 |
|
810 |
case '-': |
811 |
case '0': case '1': case '2': case '3': case '4': |
812 |
case '5': case '6': case '7': case '8': case '9': |
813 |
return decode_num (dec); |
814 |
|
815 |
case 't': |
816 |
if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
817 |
{ |
818 |
dec->cur += 4; |
819 |
return newSViv (1); |
820 |
} |
821 |
else |
822 |
ERR ("'true' expected"); |
823 |
|
824 |
break; |
825 |
|
826 |
case 'f': |
827 |
if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
828 |
{ |
829 |
dec->cur += 5; |
830 |
return newSViv (0); |
831 |
} |
832 |
else |
833 |
ERR ("'false' expected"); |
834 |
|
835 |
break; |
836 |
|
837 |
case 'n': |
838 |
if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
839 |
{ |
840 |
dec->cur += 4; |
841 |
return newSVsv (&PL_sv_undef); |
842 |
} |
843 |
else |
844 |
ERR ("'null' expected"); |
845 |
|
846 |
break; |
847 |
|
848 |
default: |
849 |
ERR ("malformed json string, neither array, object, number, string or atom"); |
850 |
break; |
851 |
} |
852 |
|
853 |
fail: |
854 |
return 0; |
855 |
} |
856 |
|
857 |
static SV * |
858 |
decode_json (SV *string, UV flags) |
859 |
{ |
860 |
SV *sv; |
861 |
|
862 |
if (flags & F_UTF8) |
863 |
sv_utf8_downgrade (string, 0); |
864 |
else |
865 |
sv_utf8_upgrade (string); |
866 |
|
867 |
SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
868 |
|
869 |
dec_t dec; |
870 |
dec.flags = flags; |
871 |
dec.cur = SvPVX (string); |
872 |
dec.end = SvEND (string); |
873 |
dec.err = 0; |
874 |
|
875 |
sv = decode_sv (&dec); |
876 |
|
877 |
if (!sv) |
878 |
{ |
879 |
IV offset = dec.flags & F_UTF8 |
880 |
? dec.cur - SvPVX (string) |
881 |
: utf8_distance (dec.cur, SvPVX (string)); |
882 |
SV *uni = sv_newmortal (); |
883 |
// horrible hack to silence warning inside pv_uni_display |
884 |
COP cop; |
885 |
memset (&cop, 0, sizeof (cop)); |
886 |
cop.cop_warnings = pWARN_NONE; |
887 |
SAVEVPTR (PL_curcop); |
888 |
PL_curcop = &cop; |
889 |
|
890 |
pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
891 |
croak ("%s, at character offset %d (%s)", |
892 |
dec.err, |
893 |
(int)offset, |
894 |
dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
895 |
} |
896 |
|
897 |
sv = sv_2mortal (sv); |
898 |
|
899 |
if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
900 |
croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); |
901 |
|
902 |
return sv; |
903 |
} |
904 |
|
905 |
MODULE = JSON::XS PACKAGE = JSON::XS |
906 |
|
907 |
BOOT: |
908 |
{ |
909 |
int i; |
910 |
|
911 |
memset (decode_hexdigit, 0xff, 256); |
912 |
for (i = 10; i--; ) |
913 |
decode_hexdigit ['0' + i] = i; |
914 |
|
915 |
for (i = 7; i--; ) |
916 |
{ |
917 |
decode_hexdigit ['a' + i] = 10 + i; |
918 |
decode_hexdigit ['A' + i] = 10 + i; |
919 |
} |
920 |
|
921 |
json_stash = gv_stashpv ("JSON::XS", 1); |
922 |
} |
923 |
|
924 |
PROTOTYPES: DISABLE |
925 |
|
926 |
SV *new (char *dummy) |
927 |
CODE: |
928 |
RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); |
929 |
OUTPUT: |
930 |
RETVAL |
931 |
|
932 |
SV *ascii (SV *self, int enable = 1) |
933 |
ALIAS: |
934 |
ascii = F_ASCII |
935 |
utf8 = F_UTF8 |
936 |
indent = F_INDENT |
937 |
canonical = F_CANONICAL |
938 |
space_before = F_SPACE_BEFORE |
939 |
space_after = F_SPACE_AFTER |
940 |
json_rpc = F_JSON_RPC |
941 |
pretty = F_PRETTY |
942 |
allow_nonref = F_ALLOW_NONREF |
943 |
shrink = F_SHRINK |
944 |
CODE: |
945 |
{ |
946 |
UV *uv = SvJSON (self); |
947 |
if (enable) |
948 |
*uv |= ix; |
949 |
else |
950 |
*uv &= ~ix; |
951 |
|
952 |
RETVAL = newSVsv (self); |
953 |
} |
954 |
OUTPUT: |
955 |
RETVAL |
956 |
|
957 |
void encode (SV *self, SV *scalar) |
958 |
PPCODE: |
959 |
XPUSHs (encode_json (scalar, *SvJSON (self))); |
960 |
|
961 |
void decode (SV *self, SV *jsonstr) |
962 |
PPCODE: |
963 |
XPUSHs (decode_json (jsonstr, *SvJSON (self))); |
964 |
|
965 |
PROTOTYPES: ENABLE |
966 |
|
967 |
void to_json (SV *scalar) |
968 |
PPCODE: |
969 |
XPUSHs (encode_json (scalar, F_UTF8)); |
970 |
|
971 |
void from_json (SV *jsonstr) |
972 |
PPCODE: |
973 |
XPUSHs (decode_json (jsonstr, F_UTF8)); |
974 |
|