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 0x00000001UL |
10 |
#define F_UTF8 0x00000002UL |
11 |
#define F_INDENT 0x00000004UL |
12 |
#define F_CANONICAL 0x00000008UL |
13 |
#define F_SPACE_BEFORE 0x00000010UL |
14 |
#define F_SPACE_AFTER 0x00000020UL |
15 |
#define F_ALLOW_NONREF 0x00000080UL |
16 |
#define F_SHRINK 0x00000100UL |
17 |
#define F_MAXDEPTH 0xf8000000UL |
18 |
#define S_MAXDEPTH 27 |
19 |
|
20 |
#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
21 |
|
22 |
// F_SELFCONVERT? <=> to_json/toJson |
23 |
// F_BLESSED? <=> { $__class__$ => } |
24 |
|
25 |
#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
26 |
#define F_DEFAULT (9UL << S_MAXDEPTH) |
27 |
|
28 |
#define INIT_SIZE 32 // initial scalar size to be allocated |
29 |
#define INDENT_STEP 3 // spaces per indentation level |
30 |
|
31 |
#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character |
32 |
#define SHORT_STRING_LEN 512 // special-case strings of up to this size |
33 |
|
34 |
#define SB do { |
35 |
#define SE } while (0) |
36 |
|
37 |
static HV *json_stash; // JSON::XS:: |
38 |
|
39 |
///////////////////////////////////////////////////////////////////////////// |
40 |
// utility functions |
41 |
|
42 |
static UV * |
43 |
SvJSON (SV *sv) |
44 |
{ |
45 |
if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) |
46 |
croak ("object is not of type JSON::XS"); |
47 |
|
48 |
return &SvUVX (SvRV (sv)); |
49 |
} |
50 |
|
51 |
static void |
52 |
shrink (SV *sv) |
53 |
{ |
54 |
sv_utf8_downgrade (sv, 1); |
55 |
if (SvLEN (sv) > SvCUR (sv) + 1) |
56 |
{ |
57 |
#ifdef SvPV_shrink_to_cur |
58 |
SvPV_shrink_to_cur (sv); |
59 |
#elif defined (SvPV_renew) |
60 |
SvPV_renew (sv, SvCUR (sv) + 1); |
61 |
#endif |
62 |
} |
63 |
} |
64 |
|
65 |
// decode an utf-8 character and return it, or (UV)-1 in |
66 |
// case of an error. |
67 |
// we special-case "safe" characters from U+80 .. U+7FF, |
68 |
// but use the very good perl function to parse anything else. |
69 |
// note that we never call this function for a ascii codepoints |
70 |
static UV |
71 |
decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
72 |
{ |
73 |
if (s[0] > 0xdf || s[0] < 0xc2) |
74 |
return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
75 |
else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
76 |
{ |
77 |
*clen = 2; |
78 |
return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
79 |
} |
80 |
else |
81 |
{ |
82 |
*clen = (STRLEN)-1; |
83 |
return (UV)-1; |
84 |
} |
85 |
} |
86 |
|
87 |
///////////////////////////////////////////////////////////////////////////// |
88 |
// encoder |
89 |
|
90 |
// structure used for encoding JSON |
91 |
typedef struct |
92 |
{ |
93 |
char *cur; // SvPVX (sv) + current output position |
94 |
char *end; // SvEND (sv) |
95 |
SV *sv; // result scalar |
96 |
U32 flags; // F_* |
97 |
U32 indent; // indentation level |
98 |
U32 maxdepth; // max. indentation/recursion level |
99 |
} enc_t; |
100 |
|
101 |
static void |
102 |
need (enc_t *enc, STRLEN len) |
103 |
{ |
104 |
if (enc->cur + len >= enc->end) |
105 |
{ |
106 |
STRLEN cur = enc->cur - SvPVX (enc->sv); |
107 |
SvGROW (enc->sv, cur + len + 1); |
108 |
enc->cur = SvPVX (enc->sv) + cur; |
109 |
enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); |
110 |
} |
111 |
} |
112 |
|
113 |
static void |
114 |
encode_ch (enc_t *enc, char ch) |
115 |
{ |
116 |
need (enc, 1); |
117 |
*enc->cur++ = ch; |
118 |
} |
119 |
|
120 |
static void |
121 |
encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) |
122 |
{ |
123 |
char *end = str + len; |
124 |
|
125 |
need (enc, len); |
126 |
|
127 |
while (str < end) |
128 |
{ |
129 |
unsigned char ch = *(unsigned char *)str; |
130 |
|
131 |
if (ch >= 0x20 && ch < 0x80) // most common case |
132 |
{ |
133 |
if (ch == '"') // but with slow exceptions |
134 |
{ |
135 |
need (enc, len += 1); |
136 |
*enc->cur++ = '\\'; |
137 |
*enc->cur++ = '"'; |
138 |
} |
139 |
else if (ch == '\\') |
140 |
{ |
141 |
need (enc, len += 1); |
142 |
*enc->cur++ = '\\'; |
143 |
*enc->cur++ = '\\'; |
144 |
} |
145 |
else |
146 |
*enc->cur++ = ch; |
147 |
|
148 |
++str; |
149 |
} |
150 |
else |
151 |
{ |
152 |
switch (ch) |
153 |
{ |
154 |
case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
155 |
case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
156 |
case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
157 |
case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
158 |
case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
159 |
|
160 |
default: |
161 |
{ |
162 |
STRLEN clen; |
163 |
UV uch; |
164 |
|
165 |
if (is_utf8) |
166 |
{ |
167 |
//uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
168 |
uch = decode_utf8 (str, end - str, &clen); |
169 |
if (clen == (STRLEN)-1) |
170 |
croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
171 |
} |
172 |
else |
173 |
{ |
174 |
uch = ch; |
175 |
clen = 1; |
176 |
} |
177 |
|
178 |
if (uch > 0x10FFFFUL) |
179 |
croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
180 |
|
181 |
if (uch < 0x80 || enc->flags & F_ASCII) |
182 |
{ |
183 |
if (uch > 0xFFFFUL) |
184 |
{ |
185 |
need (enc, len += 11); |
186 |
sprintf (enc->cur, "\\u%04x\\u%04x", |
187 |
(int)((uch - 0x10000) / 0x400 + 0xD800), |
188 |
(int)((uch - 0x10000) % 0x400 + 0xDC00)); |
189 |
enc->cur += 12; |
190 |
} |
191 |
else |
192 |
{ |
193 |
static char hexdigit [16] = "0123456789abcdef"; |
194 |
need (enc, len += 5); |
195 |
*enc->cur++ = '\\'; |
196 |
*enc->cur++ = 'u'; |
197 |
*enc->cur++ = hexdigit [ uch >> 12 ]; |
198 |
*enc->cur++ = hexdigit [(uch >> 8) & 15]; |
199 |
*enc->cur++ = hexdigit [(uch >> 4) & 15]; |
200 |
*enc->cur++ = hexdigit [(uch >> 0) & 15]; |
201 |
} |
202 |
|
203 |
str += clen; |
204 |
} |
205 |
else if (is_utf8) |
206 |
{ |
207 |
need (enc, len += clen); |
208 |
do |
209 |
{ |
210 |
*enc->cur++ = *str++; |
211 |
} |
212 |
while (--clen); |
213 |
} |
214 |
else |
215 |
{ |
216 |
need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed |
217 |
enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
218 |
++str; |
219 |
} |
220 |
} |
221 |
} |
222 |
} |
223 |
|
224 |
--len; |
225 |
} |
226 |
} |
227 |
|
228 |
static void |
229 |
encode_indent (enc_t *enc) |
230 |
{ |
231 |
if (enc->flags & F_INDENT) |
232 |
{ |
233 |
int spaces = enc->indent * INDENT_STEP; |
234 |
|
235 |
need (enc, spaces); |
236 |
memset (enc->cur, ' ', spaces); |
237 |
enc->cur += spaces; |
238 |
} |
239 |
} |
240 |
|
241 |
static void |
242 |
encode_space (enc_t *enc) |
243 |
{ |
244 |
need (enc, 1); |
245 |
encode_ch (enc, ' '); |
246 |
} |
247 |
|
248 |
static void |
249 |
encode_nl (enc_t *enc) |
250 |
{ |
251 |
if (enc->flags & F_INDENT) |
252 |
{ |
253 |
need (enc, 1); |
254 |
encode_ch (enc, '\n'); |
255 |
} |
256 |
} |
257 |
|
258 |
static void |
259 |
encode_comma (enc_t *enc) |
260 |
{ |
261 |
encode_ch (enc, ','); |
262 |
|
263 |
if (enc->flags & F_INDENT) |
264 |
encode_nl (enc); |
265 |
else if (enc->flags & F_SPACE_AFTER) |
266 |
encode_space (enc); |
267 |
} |
268 |
|
269 |
static void encode_sv (enc_t *enc, SV *sv); |
270 |
|
271 |
static void |
272 |
encode_av (enc_t *enc, AV *av) |
273 |
{ |
274 |
int i, len = av_len (av); |
275 |
|
276 |
if (enc->indent >= enc->maxdepth) |
277 |
croak ("data structure too deep (hit recursion limit)"); |
278 |
|
279 |
encode_ch (enc, '['); encode_nl (enc); |
280 |
++enc->indent; |
281 |
|
282 |
for (i = 0; i <= len; ++i) |
283 |
{ |
284 |
encode_indent (enc); |
285 |
encode_sv (enc, *av_fetch (av, i, 0)); |
286 |
|
287 |
if (i < len) |
288 |
encode_comma (enc); |
289 |
} |
290 |
|
291 |
encode_nl (enc); |
292 |
|
293 |
--enc->indent; |
294 |
encode_indent (enc); encode_ch (enc, ']'); |
295 |
} |
296 |
|
297 |
static void |
298 |
encode_he (enc_t *enc, HE *he) |
299 |
{ |
300 |
encode_ch (enc, '"'); |
301 |
|
302 |
if (HeKLEN (he) == HEf_SVKEY) |
303 |
{ |
304 |
SV *sv = HeSVKEY (he); |
305 |
STRLEN len; |
306 |
char *str; |
307 |
|
308 |
SvGETMAGIC (sv); |
309 |
str = SvPV (sv, len); |
310 |
|
311 |
encode_str (enc, str, len, SvUTF8 (sv)); |
312 |
} |
313 |
else |
314 |
encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
315 |
|
316 |
encode_ch (enc, '"'); |
317 |
|
318 |
if (enc->flags & F_SPACE_BEFORE) encode_space (enc); |
319 |
encode_ch (enc, ':'); |
320 |
if (enc->flags & F_SPACE_AFTER ) encode_space (enc); |
321 |
encode_sv (enc, HeVAL (he)); |
322 |
} |
323 |
|
324 |
// compare hash entries, used when all keys are bytestrings |
325 |
static int |
326 |
he_cmp_fast (const void *a_, const void *b_) |
327 |
{ |
328 |
int cmp; |
329 |
|
330 |
HE *a = *(HE **)a_; |
331 |
HE *b = *(HE **)b_; |
332 |
|
333 |
STRLEN la = HeKLEN (a); |
334 |
STRLEN lb = HeKLEN (b); |
335 |
|
336 |
if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) |
337 |
cmp = la - lb; |
338 |
|
339 |
return cmp; |
340 |
} |
341 |
|
342 |
// compare hash entries, used when some keys are sv's or utf-x |
343 |
static int |
344 |
he_cmp_slow (const void *a, const void *b) |
345 |
{ |
346 |
return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); |
347 |
} |
348 |
|
349 |
static void |
350 |
encode_hv (enc_t *enc, HV *hv) |
351 |
{ |
352 |
int count, i; |
353 |
|
354 |
if (enc->indent >= enc->maxdepth) |
355 |
croak ("data structure too deep (hit recursion limit)"); |
356 |
|
357 |
encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; |
358 |
|
359 |
if ((count = hv_iterinit (hv))) |
360 |
{ |
361 |
// for canonical output we have to sort by keys first |
362 |
// actually, this is mostly due to the stupid so-called |
363 |
// security workaround added somewhere in 5.8.x. |
364 |
// that randomises hash orderings |
365 |
if (enc->flags & F_CANONICAL) |
366 |
{ |
367 |
HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode |
368 |
int fast = 1; |
369 |
|
370 |
i = 0; |
371 |
while ((he = hv_iternext (hv))) |
372 |
{ |
373 |
hes [i++] = he; |
374 |
if (HeKLEN (he) < 0 || HeKUTF8 (he)) |
375 |
fast = 0; |
376 |
} |
377 |
|
378 |
assert (i == count); |
379 |
|
380 |
if (fast) |
381 |
qsort (hes, count, sizeof (HE *), he_cmp_fast); |
382 |
else |
383 |
{ |
384 |
// hack to forcefully disable "use bytes" |
385 |
COP cop = *PL_curcop; |
386 |
cop.op_private = 0; |
387 |
|
388 |
ENTER; |
389 |
SAVETMPS; |
390 |
|
391 |
SAVEVPTR (PL_curcop); |
392 |
PL_curcop = &cop; |
393 |
|
394 |
qsort (hes, count, sizeof (HE *), he_cmp_slow); |
395 |
|
396 |
FREETMPS; |
397 |
LEAVE; |
398 |
} |
399 |
|
400 |
for (i = 0; i < count; ++i) |
401 |
{ |
402 |
encode_indent (enc); |
403 |
encode_he (enc, hes [i]); |
404 |
|
405 |
if (i < count - 1) |
406 |
encode_comma (enc); |
407 |
} |
408 |
|
409 |
encode_nl (enc); |
410 |
} |
411 |
else |
412 |
{ |
413 |
HE *he = hv_iternext (hv); |
414 |
|
415 |
for (;;) |
416 |
{ |
417 |
encode_indent (enc); |
418 |
encode_he (enc, he); |
419 |
|
420 |
if (!(he = hv_iternext (hv))) |
421 |
break; |
422 |
|
423 |
encode_comma (enc); |
424 |
} |
425 |
|
426 |
encode_nl (enc); |
427 |
} |
428 |
} |
429 |
|
430 |
--enc->indent; encode_indent (enc); encode_ch (enc, '}'); |
431 |
} |
432 |
|
433 |
// encode objects, arrays and special \0=false and \1=true values. |
434 |
static void |
435 |
encode_rv (enc_t *enc, SV *sv) |
436 |
{ |
437 |
svtype svt; |
438 |
|
439 |
SvGETMAGIC (sv); |
440 |
svt = SvTYPE (sv); |
441 |
|
442 |
if (svt == SVt_PVHV) |
443 |
encode_hv (enc, (HV *)sv); |
444 |
else if (svt == SVt_PVAV) |
445 |
encode_av (enc, (AV *)sv); |
446 |
else if (svt < SVt_PVAV) |
447 |
{ |
448 |
if (SvNIOK (sv) && SvIV (sv) == 0) |
449 |
encode_str (enc, "false", 5, 0); |
450 |
else if (SvNIOK (sv) && SvIV (sv) == 1) |
451 |
encode_str (enc, "true", 4, 0); |
452 |
else |
453 |
croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
454 |
SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
455 |
} |
456 |
else |
457 |
croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
458 |
SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
459 |
} |
460 |
|
461 |
static void |
462 |
encode_sv (enc_t *enc, SV *sv) |
463 |
{ |
464 |
SvGETMAGIC (sv); |
465 |
|
466 |
if (SvPOKp (sv)) |
467 |
{ |
468 |
STRLEN len; |
469 |
char *str = SvPV (sv, len); |
470 |
encode_ch (enc, '"'); |
471 |
encode_str (enc, str, len, SvUTF8 (sv)); |
472 |
encode_ch (enc, '"'); |
473 |
} |
474 |
else if (SvNOKp (sv)) |
475 |
{ |
476 |
need (enc, NV_DIG + 32); |
477 |
Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
478 |
enc->cur += strlen (enc->cur); |
479 |
} |
480 |
else if (SvIOKp (sv)) |
481 |
{ |
482 |
need (enc, 64); |
483 |
enc->cur += |
484 |
SvIsUV(sv) |
485 |
? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) |
486 |
: snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); |
487 |
} |
488 |
else if (SvROK (sv)) |
489 |
encode_rv (enc, SvRV (sv)); |
490 |
else if (!SvOK (sv)) |
491 |
encode_str (enc, "null", 4, 0); |
492 |
else |
493 |
croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
494 |
SvPV_nolen (sv), SvFLAGS (sv)); |
495 |
} |
496 |
|
497 |
static SV * |
498 |
encode_json (SV *scalar, U32 flags) |
499 |
{ |
500 |
enc_t enc; |
501 |
|
502 |
if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
503 |
croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
504 |
|
505 |
enc.flags = flags; |
506 |
enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
507 |
enc.cur = SvPVX (enc.sv); |
508 |
enc.end = SvEND (enc.sv); |
509 |
enc.indent = 0; |
510 |
enc.maxdepth = DEC_DEPTH (flags); |
511 |
|
512 |
SvPOK_only (enc.sv); |
513 |
encode_sv (&enc, scalar); |
514 |
|
515 |
if (!(flags & (F_ASCII | F_UTF8))) |
516 |
SvUTF8_on (enc.sv); |
517 |
|
518 |
SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
519 |
|
520 |
if (enc.flags & F_SHRINK) |
521 |
shrink (enc.sv); |
522 |
|
523 |
return enc.sv; |
524 |
} |
525 |
|
526 |
///////////////////////////////////////////////////////////////////////////// |
527 |
// decoder |
528 |
|
529 |
// structure used for decoding JSON |
530 |
typedef struct |
531 |
{ |
532 |
char *cur; // current parser pointer |
533 |
char *end; // end of input string |
534 |
const char *err; // parse error, if != 0 |
535 |
U32 flags; // F_* |
536 |
U32 depth; // recursion depth |
537 |
U32 maxdepth; // recursion depth limit |
538 |
} dec_t; |
539 |
|
540 |
static void |
541 |
decode_ws (dec_t *dec) |
542 |
{ |
543 |
for (;;) |
544 |
{ |
545 |
char ch = *dec->cur; |
546 |
|
547 |
if (ch > 0x20 |
548 |
|| (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) |
549 |
break; |
550 |
|
551 |
++dec->cur; |
552 |
} |
553 |
} |
554 |
|
555 |
#define ERR(reason) SB dec->err = reason; goto fail; SE |
556 |
|
557 |
#define EXPECT_CH(ch) SB \ |
558 |
if (*dec->cur != ch) \ |
559 |
ERR (# ch " expected"); \ |
560 |
++dec->cur; \ |
561 |
SE |
562 |
|
563 |
#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") |
564 |
#define DEC_DEC_DEPTH --dec->depth |
565 |
|
566 |
static SV *decode_sv (dec_t *dec); |
567 |
|
568 |
static signed char decode_hexdigit[256]; |
569 |
|
570 |
static UV |
571 |
decode_4hex (dec_t *dec) |
572 |
{ |
573 |
signed char d1, d2, d3, d4; |
574 |
unsigned char *cur = (unsigned char *)dec->cur; |
575 |
|
576 |
d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); |
577 |
d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); |
578 |
d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); |
579 |
d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); |
580 |
|
581 |
dec->cur += 4; |
582 |
|
583 |
return ((UV)d1) << 12 |
584 |
| ((UV)d2) << 8 |
585 |
| ((UV)d3) << 4 |
586 |
| ((UV)d4); |
587 |
|
588 |
fail: |
589 |
return (UV)-1; |
590 |
} |
591 |
|
592 |
static SV * |
593 |
decode_str (dec_t *dec) |
594 |
{ |
595 |
SV *sv = 0; |
596 |
int utf8 = 0; |
597 |
|
598 |
do |
599 |
{ |
600 |
char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; |
601 |
char *cur = buf; |
602 |
|
603 |
do |
604 |
{ |
605 |
unsigned char ch = *(unsigned char *)dec->cur++; |
606 |
|
607 |
if (ch == '"') |
608 |
{ |
609 |
--dec->cur; |
610 |
break; |
611 |
} |
612 |
else if (ch == '\\') |
613 |
{ |
614 |
switch (*dec->cur) |
615 |
{ |
616 |
case '\\': |
617 |
case '/': |
618 |
case '"': *cur++ = *dec->cur++; break; |
619 |
|
620 |
case 'b': ++dec->cur; *cur++ = '\010'; break; |
621 |
case 't': ++dec->cur; *cur++ = '\011'; break; |
622 |
case 'n': ++dec->cur; *cur++ = '\012'; break; |
623 |
case 'f': ++dec->cur; *cur++ = '\014'; break; |
624 |
case 'r': ++dec->cur; *cur++ = '\015'; break; |
625 |
|
626 |
case 'u': |
627 |
{ |
628 |
UV lo, hi; |
629 |
++dec->cur; |
630 |
|
631 |
hi = decode_4hex (dec); |
632 |
if (hi == (UV)-1) |
633 |
goto fail; |
634 |
|
635 |
// possibly a surrogate pair |
636 |
if (hi >= 0xd800) |
637 |
if (hi < 0xdc00) |
638 |
{ |
639 |
if (dec->cur [0] != '\\' || dec->cur [1] != 'u') |
640 |
ERR ("missing low surrogate character in surrogate pair"); |
641 |
|
642 |
dec->cur += 2; |
643 |
|
644 |
lo = decode_4hex (dec); |
645 |
if (lo == (UV)-1) |
646 |
goto fail; |
647 |
|
648 |
if (lo < 0xdc00 || lo >= 0xe000) |
649 |
ERR ("surrogate pair expected"); |
650 |
|
651 |
hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; |
652 |
} |
653 |
else if (hi < 0xe000) |
654 |
ERR ("missing high surrogate character in surrogate pair"); |
655 |
|
656 |
if (hi >= 0x80) |
657 |
{ |
658 |
utf8 = 1; |
659 |
|
660 |
cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); |
661 |
} |
662 |
else |
663 |
*cur++ = hi; |
664 |
} |
665 |
break; |
666 |
|
667 |
default: |
668 |
--dec->cur; |
669 |
ERR ("illegal backslash escape sequence in string"); |
670 |
} |
671 |
} |
672 |
else if (ch >= 0x20 && ch <= 0x7f) |
673 |
*cur++ = ch; |
674 |
else if (ch >= 0x80) |
675 |
{ |
676 |
STRLEN clen; |
677 |
UV uch; |
678 |
|
679 |
--dec->cur; |
680 |
|
681 |
uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); |
682 |
if (clen == (STRLEN)-1) |
683 |
ERR ("malformed UTF-8 character in JSON string"); |
684 |
|
685 |
do |
686 |
*cur++ = *dec->cur++; |
687 |
while (--clen); |
688 |
|
689 |
utf8 = 1; |
690 |
} |
691 |
else |
692 |
{ |
693 |
--dec->cur; |
694 |
|
695 |
if (!ch) |
696 |
ERR ("unexpected end of string while parsing JSON string"); |
697 |
else |
698 |
ERR ("invalid character encountered while parsing JSON string"); |
699 |
} |
700 |
} |
701 |
while (cur < buf + SHORT_STRING_LEN); |
702 |
|
703 |
{ |
704 |
STRLEN len = cur - buf; |
705 |
|
706 |
if (sv) |
707 |
{ |
708 |
SvGROW (sv, SvCUR (sv) + len + 1); |
709 |
memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
710 |
SvCUR_set (sv, SvCUR (sv) + len); |
711 |
} |
712 |
else |
713 |
sv = newSVpvn (buf, len); |
714 |
} |
715 |
} |
716 |
while (*dec->cur != '"'); |
717 |
|
718 |
++dec->cur; |
719 |
|
720 |
if (sv) |
721 |
{ |
722 |
SvPOK_only (sv); |
723 |
*SvEND (sv) = 0; |
724 |
|
725 |
if (utf8) |
726 |
SvUTF8_on (sv); |
727 |
} |
728 |
else |
729 |
sv = newSVpvn ("", 0); |
730 |
|
731 |
return sv; |
732 |
|
733 |
fail: |
734 |
return 0; |
735 |
} |
736 |
|
737 |
static SV * |
738 |
decode_num (dec_t *dec) |
739 |
{ |
740 |
int is_nv = 0; |
741 |
char *start = dec->cur; |
742 |
|
743 |
// [minus] |
744 |
if (*dec->cur == '-') |
745 |
++dec->cur; |
746 |
|
747 |
if (*dec->cur == '0') |
748 |
{ |
749 |
++dec->cur; |
750 |
if (*dec->cur >= '0' && *dec->cur <= '9') |
751 |
ERR ("malformed number (leading zero must not be followed by another digit)"); |
752 |
} |
753 |
else if (*dec->cur < '0' || *dec->cur > '9') |
754 |
ERR ("malformed number (no digits after initial minus)"); |
755 |
else |
756 |
do |
757 |
{ |
758 |
++dec->cur; |
759 |
} |
760 |
while (*dec->cur >= '0' && *dec->cur <= '9'); |
761 |
|
762 |
// [frac] |
763 |
if (*dec->cur == '.') |
764 |
{ |
765 |
++dec->cur; |
766 |
|
767 |
if (*dec->cur < '0' || *dec->cur > '9') |
768 |
ERR ("malformed number (no digits after decimal point)"); |
769 |
|
770 |
do |
771 |
{ |
772 |
++dec->cur; |
773 |
} |
774 |
while (*dec->cur >= '0' && *dec->cur <= '9'); |
775 |
|
776 |
is_nv = 1; |
777 |
} |
778 |
|
779 |
// [exp] |
780 |
if (*dec->cur == 'e' || *dec->cur == 'E') |
781 |
{ |
782 |
++dec->cur; |
783 |
|
784 |
if (*dec->cur == '-' || *dec->cur == '+') |
785 |
++dec->cur; |
786 |
|
787 |
if (*dec->cur < '0' || *dec->cur > '9') |
788 |
ERR ("malformed number (no digits after exp sign)"); |
789 |
|
790 |
do |
791 |
{ |
792 |
++dec->cur; |
793 |
} |
794 |
while (*dec->cur >= '0' && *dec->cur <= '9'); |
795 |
|
796 |
is_nv = 1; |
797 |
} |
798 |
|
799 |
if (!is_nv) |
800 |
{ |
801 |
UV uv; |
802 |
int numtype = grok_number (start, dec->cur - start, &uv); |
803 |
if (numtype & IS_NUMBER_IN_UV) |
804 |
if (numtype & IS_NUMBER_NEG) |
805 |
{ |
806 |
if (uv < (UV)IV_MIN) |
807 |
return newSViv (-(IV)uv); |
808 |
} |
809 |
else |
810 |
return newSVuv (uv); |
811 |
} |
812 |
|
813 |
return newSVnv (Atof (start)); |
814 |
|
815 |
fail: |
816 |
return 0; |
817 |
} |
818 |
|
819 |
static SV * |
820 |
decode_av (dec_t *dec) |
821 |
{ |
822 |
AV *av = newAV (); |
823 |
|
824 |
DEC_INC_DEPTH; |
825 |
decode_ws (dec); |
826 |
|
827 |
if (*dec->cur == ']') |
828 |
++dec->cur; |
829 |
else |
830 |
for (;;) |
831 |
{ |
832 |
SV *value; |
833 |
|
834 |
value = decode_sv (dec); |
835 |
if (!value) |
836 |
goto fail; |
837 |
|
838 |
av_push (av, value); |
839 |
|
840 |
decode_ws (dec); |
841 |
|
842 |
if (*dec->cur == ']') |
843 |
{ |
844 |
++dec->cur; |
845 |
break; |
846 |
} |
847 |
|
848 |
if (*dec->cur != ',') |
849 |
ERR (", or ] expected while parsing array"); |
850 |
|
851 |
++dec->cur; |
852 |
} |
853 |
|
854 |
DEC_DEC_DEPTH; |
855 |
return newRV_noinc ((SV *)av); |
856 |
|
857 |
fail: |
858 |
SvREFCNT_dec (av); |
859 |
DEC_DEC_DEPTH; |
860 |
return 0; |
861 |
} |
862 |
|
863 |
static SV * |
864 |
decode_hv (dec_t *dec) |
865 |
{ |
866 |
HV *hv = newHV (); |
867 |
|
868 |
DEC_INC_DEPTH; |
869 |
decode_ws (dec); |
870 |
|
871 |
if (*dec->cur == '}') |
872 |
++dec->cur; |
873 |
else |
874 |
for (;;) |
875 |
{ |
876 |
SV *key, *value; |
877 |
|
878 |
decode_ws (dec); EXPECT_CH ('"'); |
879 |
|
880 |
key = decode_str (dec); |
881 |
if (!key) |
882 |
goto fail; |
883 |
|
884 |
decode_ws (dec); EXPECT_CH (':'); |
885 |
|
886 |
value = decode_sv (dec); |
887 |
if (!value) |
888 |
{ |
889 |
SvREFCNT_dec (key); |
890 |
goto fail; |
891 |
} |
892 |
|
893 |
hv_store_ent (hv, key, value, 0); |
894 |
SvREFCNT_dec (key); |
895 |
|
896 |
decode_ws (dec); |
897 |
|
898 |
if (*dec->cur == '}') |
899 |
{ |
900 |
++dec->cur; |
901 |
break; |
902 |
} |
903 |
|
904 |
if (*dec->cur != ',') |
905 |
ERR (", or } expected while parsing object/hash"); |
906 |
|
907 |
++dec->cur; |
908 |
} |
909 |
|
910 |
DEC_DEC_DEPTH; |
911 |
return newRV_noinc ((SV *)hv); |
912 |
|
913 |
fail: |
914 |
SvREFCNT_dec (hv); |
915 |
DEC_DEC_DEPTH; |
916 |
return 0; |
917 |
} |
918 |
|
919 |
static SV * |
920 |
decode_sv (dec_t *dec) |
921 |
{ |
922 |
decode_ws (dec); |
923 |
switch (*dec->cur) |
924 |
{ |
925 |
case '"': ++dec->cur; return decode_str (dec); |
926 |
case '[': ++dec->cur; return decode_av (dec); |
927 |
case '{': ++dec->cur; return decode_hv (dec); |
928 |
|
929 |
case '-': |
930 |
case '0': case '1': case '2': case '3': case '4': |
931 |
case '5': case '6': case '7': case '8': case '9': |
932 |
return decode_num (dec); |
933 |
|
934 |
case 't': |
935 |
if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
936 |
{ |
937 |
dec->cur += 4; |
938 |
return newSViv (1); |
939 |
} |
940 |
else |
941 |
ERR ("'true' expected"); |
942 |
|
943 |
break; |
944 |
|
945 |
case 'f': |
946 |
if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
947 |
{ |
948 |
dec->cur += 5; |
949 |
return newSViv (0); |
950 |
} |
951 |
else |
952 |
ERR ("'false' expected"); |
953 |
|
954 |
break; |
955 |
|
956 |
case 'n': |
957 |
if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
958 |
{ |
959 |
dec->cur += 4; |
960 |
return newSVsv (&PL_sv_undef); |
961 |
} |
962 |
else |
963 |
ERR ("'null' expected"); |
964 |
|
965 |
break; |
966 |
|
967 |
default: |
968 |
ERR ("malformed JSON string, neither array, object, number, string or atom"); |
969 |
break; |
970 |
} |
971 |
|
972 |
fail: |
973 |
return 0; |
974 |
} |
975 |
|
976 |
static SV * |
977 |
decode_json (SV *string, U32 flags) |
978 |
{ |
979 |
dec_t dec; |
980 |
SV *sv; |
981 |
|
982 |
SvUPGRADE (string, SVt_PV); |
983 |
|
984 |
if (flags & F_UTF8) |
985 |
sv_utf8_downgrade (string, 0); |
986 |
else |
987 |
sv_utf8_upgrade (string); |
988 |
|
989 |
SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
990 |
|
991 |
dec.flags = flags; |
992 |
dec.cur = SvPVX (string); |
993 |
dec.end = SvEND (string); |
994 |
dec.err = 0; |
995 |
dec.depth = 0; |
996 |
dec.maxdepth = DEC_DEPTH (dec.flags); |
997 |
|
998 |
*dec.end = 0; // this should basically be a nop, too, but make sure its there |
999 |
sv = decode_sv (&dec); |
1000 |
|
1001 |
if (!sv) |
1002 |
{ |
1003 |
IV offset = dec.flags & F_UTF8 |
1004 |
? dec.cur - SvPVX (string) |
1005 |
: utf8_distance (dec.cur, SvPVX (string)); |
1006 |
SV *uni = sv_newmortal (); |
1007 |
|
1008 |
// horrible hack to silence warning inside pv_uni_display |
1009 |
COP cop = *PL_curcop; |
1010 |
cop.cop_warnings = pWARN_NONE; |
1011 |
ENTER; |
1012 |
SAVEVPTR (PL_curcop); |
1013 |
PL_curcop = &cop; |
1014 |
pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1015 |
LEAVE; |
1016 |
|
1017 |
croak ("%s, at character offset %d [\"%s\"]", |
1018 |
dec.err, |
1019 |
(int)offset, |
1020 |
dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1021 |
} |
1022 |
|
1023 |
sv = sv_2mortal (sv); |
1024 |
|
1025 |
if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1026 |
croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1027 |
|
1028 |
return sv; |
1029 |
} |
1030 |
|
1031 |
///////////////////////////////////////////////////////////////////////////// |
1032 |
// XS interface functions |
1033 |
|
1034 |
MODULE = JSON::XS PACKAGE = JSON::XS |
1035 |
|
1036 |
BOOT: |
1037 |
{ |
1038 |
int i; |
1039 |
|
1040 |
memset (decode_hexdigit, 0xff, 256); |
1041 |
|
1042 |
for (i = 0; i < 256; ++i) |
1043 |
decode_hexdigit [i] = |
1044 |
i >= '0' && i <= '9' ? i - '0' |
1045 |
: i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1046 |
: i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1047 |
: -1; |
1048 |
|
1049 |
json_stash = gv_stashpv ("JSON::XS", 1); |
1050 |
} |
1051 |
|
1052 |
PROTOTYPES: DISABLE |
1053 |
|
1054 |
SV *new (char *dummy) |
1055 |
CODE: |
1056 |
RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); |
1057 |
OUTPUT: |
1058 |
RETVAL |
1059 |
|
1060 |
SV *ascii (SV *self, int enable = 1) |
1061 |
ALIAS: |
1062 |
ascii = F_ASCII |
1063 |
utf8 = F_UTF8 |
1064 |
indent = F_INDENT |
1065 |
canonical = F_CANONICAL |
1066 |
space_before = F_SPACE_BEFORE |
1067 |
space_after = F_SPACE_AFTER |
1068 |
pretty = F_PRETTY |
1069 |
allow_nonref = F_ALLOW_NONREF |
1070 |
shrink = F_SHRINK |
1071 |
CODE: |
1072 |
{ |
1073 |
UV *uv = SvJSON (self); |
1074 |
if (enable) |
1075 |
*uv |= ix; |
1076 |
else |
1077 |
*uv &= ~ix; |
1078 |
|
1079 |
RETVAL = newSVsv (self); |
1080 |
} |
1081 |
OUTPUT: |
1082 |
RETVAL |
1083 |
|
1084 |
SV *max_depth (SV *self, UV max_depth = 0x80000000UL) |
1085 |
CODE: |
1086 |
{ |
1087 |
UV *uv = SvJSON (self); |
1088 |
UV log2 = 0; |
1089 |
|
1090 |
if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
1091 |
|
1092 |
while ((1UL << log2) < max_depth) |
1093 |
++log2; |
1094 |
|
1095 |
*uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
1096 |
|
1097 |
RETVAL = newSVsv (self); |
1098 |
} |
1099 |
OUTPUT: |
1100 |
RETVAL |
1101 |
|
1102 |
void encode (SV *self, SV *scalar) |
1103 |
PPCODE: |
1104 |
XPUSHs (encode_json (scalar, *SvJSON (self))); |
1105 |
|
1106 |
void decode (SV *self, SV *jsonstr) |
1107 |
PPCODE: |
1108 |
XPUSHs (decode_json (jsonstr, *SvJSON (self))); |
1109 |
|
1110 |
PROTOTYPES: ENABLE |
1111 |
|
1112 |
void to_json (SV *scalar) |
1113 |
ALIAS: |
1114 |
objToJson = 0 |
1115 |
PPCODE: |
1116 |
XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); |
1117 |
|
1118 |
void from_json (SV *jsonstr) |
1119 |
ALIAS: |
1120 |
jsonToObj = 0 |
1121 |
PPCODE: |
1122 |
XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); |
1123 |
|