ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/JSON-XS/XS.xs
(Generate patch)

Comparing JSON-XS/XS.xs (file contents):
Revision 1.55 by root, Mon Jul 23 22:57:40 2007 UTC vs.
Revision 1.121 by root, Tue Oct 29 00:06:40 2013 UTC

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 <limits.h>
10#include <float.h>
9 11
10#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 14#endif
13 15
14// some old perls do not have this, try to make it work, no 16// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces. 17// guarantees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
18#endif 20#endif
21
22// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
19 24
20#define F_ASCII 0x00000001UL 25#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL 26#define F_LATIN1 0x00000002UL
22#define F_UTF8 0x00000004UL 27#define F_UTF8 0x00000004UL
23#define F_INDENT 0x00000008UL 28#define F_INDENT 0x00000008UL
… …
26#define F_SPACE_AFTER 0x00000040UL 31#define F_SPACE_AFTER 0x00000040UL
27#define F_ALLOW_NONREF 0x00000100UL 32#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL 33#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL 34#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL 35#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL 36#define F_RELAXED 0x00001000UL
34#define S_MAXSIZE 20 37#define F_ALLOW_UNKNOWN 0x00002000UL
38#define F_ALLOW_TAGS 0x00004000UL
35#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 39#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
36 40
37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
39
40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
41#define F_DEFAULT (9UL << S_MAXDEPTH)
42 42
43#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
44#define INDENT_STEP 3 // spaces per indentation level 44#define INDENT_STEP 3 // spaces per indentation level
45 45
46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
47 47
48#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
49
48#define SB do { 50#define SB do {
49#define SE } while (0) 51#define SE } while (0)
50 52
51#if __GNUC__ >= 3 53#if __GNUC__ >= 3
52# define expect(expr,value) __builtin_expect ((expr),(value)) 54# define expect(expr,value) __builtin_expect ((expr), (value))
53# define inline inline 55# define INLINE static inline
54#else 56#else
55# define expect(expr,value) (expr) 57# define expect(expr,value) (expr)
56# define inline static 58# define INLINE static
57#endif 59#endif
58 60
59#define expect_false(expr) expect ((expr) != 0, 0) 61#define expect_false(expr) expect ((expr) != 0, 0)
60#define expect_true(expr) expect ((expr) != 0, 1) 62#define expect_true(expr) expect ((expr) != 0, 1)
61 63
64#define IN_RANGE_INC(type,val,beg,end) \
65 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
66 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
67
68#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
69
70#ifdef USE_ITHREADS
71# define JSON_SLOW 1
72# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
73#else
74# define JSON_SLOW 0
75# define JSON_STASH json_stash
76#endif
77
78// the amount of HEs to allocate on the stack, when sorting keys
79#define STACK_HES 64
80
62static HV *json_stash, *json_boolean_stash; // JSON::XS:: 81static HV *json_stash, *types_boolean_stash; // JSON::XS::
63static SV *json_true, *json_false; 82static SV *types_true, *types_false, *sv_json;
83
84enum {
85 INCR_M_WS = 0, // initial whitespace skipping, must be 0
86 INCR_M_STR, // inside string
87 INCR_M_BS, // inside backslash
88 INCR_M_C0, // inside comment in initial whitespace sequence
89 INCR_M_C1, // inside comment in other places
90 INCR_M_JSON // outside anything, count nesting
91};
92
93#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
64 94
65typedef struct { 95typedef struct {
66 U32 flags; 96 U32 flags;
97 U32 max_depth;
98 STRLEN max_size;
99
67 SV *cb_object; 100 SV *cb_object;
68 HV *cb_sk_object; 101 HV *cb_sk_object;
102
103 // for the incremental parser
104 SV *incr_text; // the source text so far
105 STRLEN incr_pos; // the current offset into the text
106 int incr_nest; // {[]}-nesting level
107 unsigned char incr_mode;
69} JSON; 108} JSON;
109
110INLINE void
111json_init (JSON *json)
112{
113 Zero (json, 1, JSON);
114 json->max_depth = 512;
115}
70 116
71///////////////////////////////////////////////////////////////////////////// 117/////////////////////////////////////////////////////////////////////////////
72// utility functions 118// utility functions
73 119
74inline void 120INLINE SV *
121get_bool (const char *name)
122{
123 SV *sv = get_sv (name, 1);
124
125 SvREADONLY_on (sv);
126 SvREADONLY_on (SvRV (sv));
127
128 return sv;
129}
130
131INLINE void
75shrink (SV *sv) 132shrink (SV *sv)
76{ 133{
77 sv_utf8_downgrade (sv, 1); 134 sv_utf8_downgrade (sv, 1);
135
78 if (SvLEN (sv) > SvCUR (sv) + 1) 136 if (SvLEN (sv) > SvCUR (sv) + 1)
79 { 137 {
80#ifdef SvPV_shrink_to_cur 138#ifdef SvPV_shrink_to_cur
81 SvPV_shrink_to_cur (sv); 139 SvPV_shrink_to_cur (sv);
82#elif defined (SvPV_renew) 140#elif defined (SvPV_renew)
… …
88// decode an utf-8 character and return it, or (UV)-1 in 146// decode an utf-8 character and return it, or (UV)-1 in
89// case of an error. 147// case of an error.
90// we special-case "safe" characters from U+80 .. U+7FF, 148// we special-case "safe" characters from U+80 .. U+7FF,
91// but use the very good perl function to parse anything else. 149// but use the very good perl function to parse anything else.
92// note that we never call this function for a ascii codepoints 150// note that we never call this function for a ascii codepoints
93inline UV 151INLINE UV
94decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 152decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
95{ 153{
96 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 154 if (expect_true (len >= 2
97 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 155 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
98 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 156 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
99 { 157 {
100 *clen = 2; 158 *clen = 2;
101 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 159 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
102 } 160 }
103 else 161 else
104 { 162 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
105 *clen = (STRLEN)-1; 163}
106 return (UV)-1; 164
165// likewise for encoding, also never called for ascii codepoints
166// this function takes advantage of this fact, although current gccs
167// seem to optimise the check for >= 0x80 away anyways
168INLINE unsigned char *
169encode_utf8 (unsigned char *s, UV ch)
170{
171 if (expect_false (ch < 0x000080))
172 *s++ = ch;
173 else if (expect_true (ch < 0x000800))
174 *s++ = 0xc0 | ( ch >> 6),
175 *s++ = 0x80 | ( ch & 0x3f);
176 else if ( ch < 0x010000)
177 *s++ = 0xe0 | ( ch >> 12),
178 *s++ = 0x80 | ((ch >> 6) & 0x3f),
179 *s++ = 0x80 | ( ch & 0x3f);
180 else if ( ch < 0x110000)
181 *s++ = 0xf0 | ( ch >> 18),
182 *s++ = 0x80 | ((ch >> 12) & 0x3f),
183 *s++ = 0x80 | ((ch >> 6) & 0x3f),
184 *s++ = 0x80 | ( ch & 0x3f);
185
186 return s;
187}
188
189// convert offset pointer to character index, sv must be string
190static STRLEN
191ptr_to_index (SV *sv, char *offset)
192{
193 return SvUTF8 (sv)
194 ? utf8_distance (offset, SvPVX (sv))
195 : offset - SvPVX (sv);
196}
197
198/////////////////////////////////////////////////////////////////////////////
199// fp hell
200
201// scan a group of digits, and a trailing exponent
202static void
203json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
204{
205 UV uaccum = 0;
206 int eaccum = 0;
207
208 // if we recurse too deep, skip all remaining digits
209 // to avoid a stack overflow attack
210 if (expect_false (--maxdepth <= 0))
211 while (((U8)*s - '0') < 10)
212 ++s;
213
214 for (;;)
107 } 215 {
108} 216 U8 dig = (U8)*s - '0';
109 217
218 if (expect_false (dig >= 10))
219 {
220 if (dig == (U8)((U8)'.' - (U8)'0'))
221 {
222 ++s;
223 json_atof_scan1 (s, accum, expo, 1, maxdepth);
224 }
225 else if ((dig | ' ') == 'e' - '0')
226 {
227 int exp2 = 0;
228 int neg = 0;
229
230 ++s;
231
232 if (*s == '-')
233 {
234 ++s;
235 neg = 1;
236 }
237 else if (*s == '+')
238 ++s;
239
240 while ((dig = (U8)*s - '0') < 10)
241 exp2 = exp2 * 10 + *s++ - '0';
242
243 *expo += neg ? -exp2 : exp2;
244 }
245
246 break;
247 }
248
249 ++s;
250
251 uaccum = uaccum * 10 + dig;
252 ++eaccum;
253
254 // if we have too many digits, then recurse for more
255 // we actually do this for rather few digits
256 if (uaccum >= (UV_MAX - 9) / 10)
257 {
258 if (postdp) *expo -= eaccum;
259 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
260 if (postdp) *expo += eaccum;
261
262 break;
263 }
264 }
265
266 // this relies greatly on the quality of the pow ()
267 // implementation of the platform, but a good
268 // implementation is hard to beat.
269 // (IEEE 754 conformant ones are required to be exact)
270 if (postdp) *expo -= eaccum;
271 *accum += uaccum * Perl_pow (10., *expo);
272 *expo += eaccum;
273}
274
275static NV
276json_atof (const char *s)
277{
278 NV accum = 0.;
279 int expo = 0;
280 int neg = 0;
281
282 if (*s == '-')
283 {
284 ++s;
285 neg = 1;
286 }
287
288 // a recursion depth of ten gives us >>500 bits
289 json_atof_scan1 (s, &accum, &expo, 0, 10);
290
291 return neg ? -accum : accum;
292}
110///////////////////////////////////////////////////////////////////////////// 293/////////////////////////////////////////////////////////////////////////////
111// encoder 294// encoder
112 295
113// structure used for encoding JSON 296// structure used for encoding JSON
114typedef struct 297typedef struct
… …
116 char *cur; // SvPVX (sv) + current output position 299 char *cur; // SvPVX (sv) + current output position
117 char *end; // SvEND (sv) 300 char *end; // SvEND (sv)
118 SV *sv; // result scalar 301 SV *sv; // result scalar
119 JSON json; 302 JSON json;
120 U32 indent; // indentation level 303 U32 indent; // indentation level
121 U32 maxdepth; // max. indentation/recursion level 304 UV limit; // escape character values >= this value when encoding
122} enc_t; 305} enc_t;
123 306
124inline void 307INLINE void
125need (enc_t *enc, STRLEN len) 308need (enc_t *enc, STRLEN len)
126{ 309{
127 if (expect_false (enc->cur + len >= enc->end)) 310 if (expect_false (enc->cur + len >= enc->end))
128 { 311 {
129 STRLEN cur = enc->cur - SvPVX (enc->sv); 312 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
130 SvGROW (enc->sv, cur + len + 1); 313 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
131 enc->cur = SvPVX (enc->sv) + cur; 314 enc->cur = SvPVX (enc->sv) + cur;
132 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 315 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
133 } 316 }
134} 317}
135 318
136inline void 319INLINE void
137encode_ch (enc_t *enc, char ch) 320encode_ch (enc_t *enc, char ch)
138{ 321{
139 need (enc, 1); 322 need (enc, 1);
140 *enc->cur++ = ch; 323 *enc->cur++ = ch;
141} 324}
… …
195 { 378 {
196 uch = ch; 379 uch = ch;
197 clen = 1; 380 clen = 1;
198 } 381 }
199 382
200 if (uch > 0x10FFFFUL) 383 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
201 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
202
203 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
204 { 384 {
205 if (uch > 0xFFFFUL) 385 if (uch >= 0x10000UL)
206 { 386 {
387 if (uch >= 0x110000UL)
388 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
389
207 need (enc, len += 11); 390 need (enc, len += 11);
208 sprintf (enc->cur, "\\u%04x\\u%04x", 391 sprintf (enc->cur, "\\u%04x\\u%04x",
209 (int)((uch - 0x10000) / 0x400 + 0xD800), 392 (int)((uch - 0x10000) / 0x400 + 0xD800),
210 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 393 (int)((uch - 0x10000) % 0x400 + 0xDC00));
211 enc->cur += 12; 394 enc->cur += 12;
212 } 395 }
213 else 396 else
214 { 397 {
215 static char hexdigit [16] = "0123456789abcdef";
216 need (enc, len += 5); 398 need (enc, len += 5);
217 *enc->cur++ = '\\'; 399 *enc->cur++ = '\\';
218 *enc->cur++ = 'u'; 400 *enc->cur++ = 'u';
219 *enc->cur++ = hexdigit [ uch >> 12 ]; 401 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
220 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 402 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
221 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 403 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
222 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 404 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
223 } 405 }
224 406
225 str += clen; 407 str += clen;
226 } 408 }
227 else if (enc->json.flags & F_LATIN1) 409 else if (enc->json.flags & F_LATIN1)
… …
239 while (--clen); 421 while (--clen);
240 } 422 }
241 else 423 else
242 { 424 {
243 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 425 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
244 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 426 enc->cur = encode_utf8 (enc->cur, uch);
245 ++str; 427 ++str;
246 } 428 }
247 } 429 }
248 } 430 }
249 } 431 }
250 432
251 --len; 433 --len;
252 } 434 }
253} 435}
254 436
255inline void 437INLINE void
256encode_indent (enc_t *enc) 438encode_indent (enc_t *enc)
257{ 439{
258 if (enc->json.flags & F_INDENT) 440 if (enc->json.flags & F_INDENT)
259 { 441 {
260 int spaces = enc->indent * INDENT_STEP; 442 int spaces = enc->indent * INDENT_STEP;
… …
263 memset (enc->cur, ' ', spaces); 445 memset (enc->cur, ' ', spaces);
264 enc->cur += spaces; 446 enc->cur += spaces;
265 } 447 }
266} 448}
267 449
268inline void 450INLINE void
269encode_space (enc_t *enc) 451encode_space (enc_t *enc)
270{ 452{
271 need (enc, 1); 453 need (enc, 1);
272 encode_ch (enc, ' '); 454 encode_ch (enc, ' ');
273} 455}
274 456
275inline void 457INLINE void
276encode_nl (enc_t *enc) 458encode_nl (enc_t *enc)
277{ 459{
278 if (enc->json.flags & F_INDENT) 460 if (enc->json.flags & F_INDENT)
279 { 461 {
280 need (enc, 1); 462 need (enc, 1);
281 encode_ch (enc, '\n'); 463 encode_ch (enc, '\n');
282 } 464 }
283} 465}
284 466
285inline void 467INLINE void
286encode_comma (enc_t *enc) 468encode_comma (enc_t *enc)
287{ 469{
288 encode_ch (enc, ','); 470 encode_ch (enc, ',');
289 471
290 if (enc->json.flags & F_INDENT) 472 if (enc->json.flags & F_INDENT)
… …
298static void 480static void
299encode_av (enc_t *enc, AV *av) 481encode_av (enc_t *enc, AV *av)
300{ 482{
301 int i, len = av_len (av); 483 int i, len = av_len (av);
302 484
303 if (enc->indent >= enc->maxdepth) 485 if (enc->indent >= enc->json.max_depth)
304 croak ("data structure too deep (hit recursion limit)"); 486 croak (ERR_NESTING_EXCEEDED);
305 487
306 encode_ch (enc, '['); encode_nl (enc); 488 encode_ch (enc, '[');
307 ++enc->indent;
308 489
490 if (len >= 0)
491 {
492 encode_nl (enc); ++enc->indent;
493
309 for (i = 0; i <= len; ++i) 494 for (i = 0; i <= len; ++i)
310 { 495 {
311 SV **svp = av_fetch (av, i, 0); 496 SV **svp = av_fetch (av, i, 0);
312 497
313 encode_indent (enc); 498 encode_indent (enc);
314 499
315 if (svp) 500 if (svp)
316 encode_sv (enc, *svp); 501 encode_sv (enc, *svp);
317 else 502 else
318 encode_str (enc, "null", 4, 0); 503 encode_str (enc, "null", 4, 0);
319 504
320 if (i < len) 505 if (i < len)
321 encode_comma (enc); 506 encode_comma (enc);
322 } 507 }
323 508
509 encode_nl (enc); --enc->indent; encode_indent (enc);
510 }
511
324 encode_nl (enc); 512 encode_ch (enc, ']');
325
326 --enc->indent;
327 encode_indent (enc); encode_ch (enc, ']');
328} 513}
329 514
330static void 515static void
331encode_he (enc_t *enc, HE *he) 516encode_hk (enc_t *enc, HE *he)
332{ 517{
333 encode_ch (enc, '"'); 518 encode_ch (enc, '"');
334 519
335 if (HeKLEN (he) == HEf_SVKEY) 520 if (HeKLEN (he) == HEf_SVKEY)
336 { 521 {
337 SV *sv = HeSVKEY (he); 522 SV *sv = HeSVKEY (he);
338 STRLEN len; 523 STRLEN len;
339 char *str; 524 char *str;
340 525
341 SvGETMAGIC (sv); 526 SvGETMAGIC (sv);
342 str = SvPV (sv, len); 527 str = SvPV (sv, len);
343 528
344 encode_str (enc, str, len, SvUTF8 (sv)); 529 encode_str (enc, str, len, SvUTF8 (sv));
345 } 530 }
… …
349 encode_ch (enc, '"'); 534 encode_ch (enc, '"');
350 535
351 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); 536 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
352 encode_ch (enc, ':'); 537 encode_ch (enc, ':');
353 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); 538 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
354 encode_sv (enc, HeVAL (he));
355} 539}
356 540
357// compare hash entries, used when all keys are bytestrings 541// compare hash entries, used when all keys are bytestrings
358static int 542static int
359he_cmp_fast (const void *a_, const void *b_) 543he_cmp_fast (const void *a_, const void *b_)
… …
364 HE *b = *(HE **)b_; 548 HE *b = *(HE **)b_;
365 549
366 STRLEN la = HeKLEN (a); 550 STRLEN la = HeKLEN (a);
367 STRLEN lb = HeKLEN (b); 551 STRLEN lb = HeKLEN (b);
368 552
369 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 553 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
370 cmp = la - lb; 554 cmp = lb - la;
371 555
372 return cmp; 556 return cmp;
373} 557}
374 558
375// compare hash entries, used when some keys are sv's or utf-x 559// compare hash entries, used when some keys are sv's or utf-x
376static int 560static int
377he_cmp_slow (const void *a, const void *b) 561he_cmp_slow (const void *a, const void *b)
378{ 562{
379 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 563 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
380} 564}
381 565
382static void 566static void
383encode_hv (enc_t *enc, HV *hv) 567encode_hv (enc_t *enc, HV *hv)
384{ 568{
385 int count, i; 569 HE *he;
386 570
387 if (enc->indent >= enc->maxdepth) 571 if (enc->indent >= enc->json.max_depth)
388 croak ("data structure too deep (hit recursion limit)"); 572 croak (ERR_NESTING_EXCEEDED);
389 573
390 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 574 encode_ch (enc, '{');
391 575
392 if ((count = hv_iterinit (hv)))
393 {
394 // for canonical output we have to sort by keys first 576 // for canonical output we have to sort by keys first
395 // actually, this is mostly due to the stupid so-called 577 // actually, this is mostly due to the stupid so-called
396 // security workaround added somewhere in 5.8.x. 578 // security workaround added somewhere in 5.8.x
397 // that randomises hash orderings 579 // that randomises hash orderings
398 if (enc->json.flags & F_CANONICAL) 580 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
581 {
582 int count = hv_iterinit (hv);
583
584 if (SvMAGICAL (hv))
399 { 585 {
586 // need to count by iterating. could improve by dynamically building the vector below
587 // but I don't care for the speed of this special case.
588 // note also that we will run into undefined behaviour when the two iterations
589 // do not result in the same count, something I might care for in some later release.
590
591 count = 0;
592 while (hv_iternext (hv))
593 ++count;
594
595 hv_iterinit (hv);
596 }
597
598 if (count)
599 {
400 int fast = 1; 600 int i, fast = 1;
401 HE *he; 601 HE *hes_stack [STACK_HES];
402#if defined(__BORLANDC__) || defined(_MSC_VER) 602 HE **hes = hes_stack;
403 HE **hes = _alloca (count * sizeof (HE)); 603
404#else 604 // allocate larger arrays on the heap
405 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 605 if (count > STACK_HES)
406#endif 606 {
607 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
608 hes = (HE **)SvPVX (sv);
609 }
407 610
408 i = 0; 611 i = 0;
409 while ((he = hv_iternext (hv))) 612 while ((he = hv_iternext (hv)))
410 { 613 {
411 hes [i++] = he; 614 hes [i++] = he;
… …
433 636
434 FREETMPS; 637 FREETMPS;
435 LEAVE; 638 LEAVE;
436 } 639 }
437 640
438 for (i = 0; i < count; ++i) 641 encode_nl (enc); ++enc->indent;
642
643 while (count--)
439 { 644 {
440 encode_indent (enc); 645 encode_indent (enc);
646 he = hes [count];
441 encode_he (enc, hes [i]); 647 encode_hk (enc, he);
648 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
442 649
443 if (i < count - 1) 650 if (count)
444 encode_comma (enc); 651 encode_comma (enc);
445 } 652 }
446 653
447 encode_nl (enc); 654 encode_nl (enc); --enc->indent; encode_indent (enc);
448 } 655 }
656 }
449 else 657 else
658 {
659 if (hv_iterinit (hv) || SvMAGICAL (hv))
660 if ((he = hv_iternext (hv)))
450 { 661 {
451 HE *he = hv_iternext (hv); 662 encode_nl (enc); ++enc->indent;
452 663
453 for (;;) 664 for (;;)
454 { 665 {
455 encode_indent (enc); 666 encode_indent (enc);
456 encode_he (enc, he); 667 encode_hk (enc, he);
668 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
457 669
458 if (!(he = hv_iternext (hv))) 670 if (!(he = hv_iternext (hv)))
459 break; 671 break;
460 672
461 encode_comma (enc); 673 encode_comma (enc);
462 } 674 }
463 675
464 encode_nl (enc); 676 encode_nl (enc); --enc->indent; encode_indent (enc);
465 } 677 }
466 } 678 }
467 679
468 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 680 encode_ch (enc, '}');
469} 681}
470 682
471// encode objects, arrays and special \0=false and \1=true values. 683// encode objects, arrays and special \0=false and \1=true values.
472static void 684static void
473encode_rv (enc_t *enc, SV *sv) 685encode_rv (enc_t *enc, SV *sv)
474{ 686{
475 svtype svt; 687 svtype svt;
688 GV *method;
476 689
477 SvGETMAGIC (sv); 690 SvGETMAGIC (sv);
478 svt = SvTYPE (sv); 691 svt = SvTYPE (sv);
479 692
480 if (expect_false (SvOBJECT (sv))) 693 if (expect_false (SvOBJECT (sv)))
481 { 694 {
482 if (SvSTASH (sv) == json_boolean_stash) 695 HV *boolean_stash = !JSON_SLOW || types_boolean_stash
696 ? types_boolean_stash
697 : gv_stashpv ("Types::Serialiser::Boolean", 1);
698 HV *stash = SvSTASH (sv);
699
700 if (stash == boolean_stash)
483 { 701 {
484 if (SvIV (sv)) 702 if (SvIV (sv))
485 encode_str (enc, "true", 4, 0); 703 encode_str (enc, "true", 4, 0);
486 else 704 else
487 encode_str (enc, "false", 5, 0); 705 encode_str (enc, "false", 5, 0);
488 } 706 }
707 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
708 {
709 int count;
710 dSP;
711
712 ENTER; SAVETMPS; PUSHMARK (SP);
713 EXTEND (SP, 2);
714 // we re-bless the reference to get overload and other niceties right
715 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
716 PUSHs (sv_json);
717
718 PUTBACK;
719 count = call_sv ((SV *)GvCV (method), G_ARRAY);
720 SPAGAIN;
721
722 // catch this surprisingly common error
723 if (SvROK (TOPs) && SvRV (TOPs) == sv)
724 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
725
726 encode_ch (enc, '(');
727 encode_ch (enc, '"');
728 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
729 encode_ch (enc, '"');
730 encode_ch (enc, ')');
731 encode_ch (enc, '[');
732
733 while (count)
734 {
735 encode_sv (enc, SP[1 - count--]);
736
737 if (count)
738 encode_ch (enc, ',');
739 }
740
741 encode_ch (enc, ']');
742
743 PUTBACK;
744
745 FREETMPS; LEAVE;
746 }
747 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
748 {
749 dSP;
750
751 ENTER; SAVETMPS; PUSHMARK (SP);
752 // we re-bless the reference to get overload and other niceties right
753 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
754
755 // calling with G_SCALAR ensures that we always get a 1 return value
756 PUTBACK;
757 call_sv ((SV *)GvCV (method), G_SCALAR);
758 SPAGAIN;
759
760 // catch this surprisingly common error
761 if (SvROK (TOPs) && SvRV (TOPs) == sv)
762 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
763
764 sv = POPs;
765 PUTBACK;
766
767 encode_sv (enc, sv);
768
769 FREETMPS; LEAVE;
770 }
771 else if (enc->json.flags & F_ALLOW_BLESSED)
772 encode_str (enc, "null", 4, 0);
489 else 773 else
490 {
491#if 0
492 if (0 && sv_derived_from (rv, "JSON::Literal"))
493 {
494 // not yet
495 }
496#endif
497 if (enc->json.flags & F_CONV_BLESSED)
498 {
499 // we re-bless the reference to get overload and other niceties right
500 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
501
502 if (to_json)
503 {
504 int count;
505 dSP;
506
507 ENTER; SAVETMPS; PUSHMARK (SP);
508 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
509
510 // calling with G_SCALAR ensures that we always get a 1 return value
511 PUTBACK;
512 call_sv ((SV *)GvCV (to_json), G_SCALAR);
513 SPAGAIN;
514
515 // catch this surprisingly common error
516 if (SvROK (TOPs) && SvRV (TOPs) == sv)
517 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
518
519 sv = POPs;
520 PUTBACK;
521
522 encode_sv (enc, sv);
523
524 FREETMPS; LEAVE;
525 }
526 else if (enc->json.flags & F_ALLOW_BLESSED)
527 encode_str (enc, "null", 4, 0);
528 else
529 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
530 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
531 }
532 else if (enc->json.flags & F_ALLOW_BLESSED)
533 encode_str (enc, "null", 4, 0);
534 else
535 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 774 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
536 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 775 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
537 }
538 } 776 }
539 else if (svt == SVt_PVHV) 777 else if (svt == SVt_PVHV)
540 encode_hv (enc, (HV *)sv); 778 encode_hv (enc, (HV *)sv);
541 else if (svt == SVt_PVAV) 779 else if (svt == SVt_PVAV)
542 encode_av (enc, (AV *)sv); 780 encode_av (enc, (AV *)sv);
… …
547 785
548 if (len == 1 && *pv == '1') 786 if (len == 1 && *pv == '1')
549 encode_str (enc, "true", 4, 0); 787 encode_str (enc, "true", 4, 0);
550 else if (len == 1 && *pv == '0') 788 else if (len == 1 && *pv == '0')
551 encode_str (enc, "false", 5, 0); 789 encode_str (enc, "false", 5, 0);
790 else if (enc->json.flags & F_ALLOW_UNKNOWN)
791 encode_str (enc, "null", 4, 0);
552 else 792 else
553 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 793 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
554 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 794 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
555 } 795 }
796 else if (enc->json.flags & F_ALLOW_UNKNOWN)
797 encode_str (enc, "null", 4, 0);
556 else 798 else
557 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 799 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
558 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 800 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
559} 801}
560 802
… …
578 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 820 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
579 enc->cur += strlen (enc->cur); 821 enc->cur += strlen (enc->cur);
580 } 822 }
581 else if (SvIOKp (sv)) 823 else if (SvIOKp (sv))
582 { 824 {
583 // we assume we can always read an IV as a UV 825 // we assume we can always read an IV as a UV and vice versa
584 if (SvUV (sv) & ~(UV)0x7fff) 826 // we assume two's complement
585 { 827 // we assume no aliasing issues in the union
586 // large integer, use the (rather slow) snprintf way. 828 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
587 need (enc, sizeof (UV) * 3); 829 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
588 enc->cur +=
589 SvIsUV(sv)
590 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
591 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
592 }
593 else
594 { 830 {
595 // optimise the "small number case" 831 // optimise the "small number case"
596 // code will likely be branchless and use only a single multiplication 832 // code will likely be branchless and use only a single multiplication
833 // works for numbers up to 59074
597 I32 i = SvIV (sv); 834 I32 i = SvIVX (sv);
598 U32 u; 835 U32 u;
599 char digit, nz = 0; 836 char digit, nz = 0;
600 837
601 need (enc, 6); 838 need (enc, 6);
602 839
… …
608 845
609 // now output digit by digit, each time masking out the integer part 846 // now output digit by digit, each time masking out the integer part
610 // and multiplying by 5 while moving the decimal point one to the right, 847 // and multiplying by 5 while moving the decimal point one to the right,
611 // resulting in a net multiplication by 10. 848 // resulting in a net multiplication by 10.
612 // we always write the digit to memory but conditionally increment 849 // we always write the digit to memory but conditionally increment
613 // the pointer, to ease the usage of conditional move instructions. 850 // the pointer, to enable the use of conditional move instructions.
614 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 851 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
615 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 852 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
616 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 853 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
617 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 854 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
618 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 855 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
619 } 856 }
857 else
858 {
859 // large integer, use the (rather slow) snprintf way.
860 need (enc, IVUV_MAXCHARS);
861 enc->cur +=
862 SvIsUV(sv)
863 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
864 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
865 }
620 } 866 }
621 else if (SvROK (sv)) 867 else if (SvROK (sv))
622 encode_rv (enc, SvRV (sv)); 868 encode_rv (enc, SvRV (sv));
623 else if (!SvOK (sv)) 869 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
624 encode_str (enc, "null", 4, 0); 870 encode_str (enc, "null", 4, 0);
625 else 871 else
626 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 872 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
627 SvPV_nolen (sv), SvFLAGS (sv)); 873 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
628} 874}
629 875
630static SV * 876static SV *
631encode_json (SV *scalar, JSON *json) 877encode_json (SV *scalar, JSON *json)
632{ 878{
… …
638 enc.json = *json; 884 enc.json = *json;
639 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 885 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
640 enc.cur = SvPVX (enc.sv); 886 enc.cur = SvPVX (enc.sv);
641 enc.end = SvEND (enc.sv); 887 enc.end = SvEND (enc.sv);
642 enc.indent = 0; 888 enc.indent = 0;
643 enc.maxdepth = DEC_DEPTH (enc.json.flags); 889 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
890 : enc.json.flags & F_LATIN1 ? 0x000100UL
891 : 0x110000UL;
644 892
645 SvPOK_only (enc.sv); 893 SvPOK_only (enc.sv);
646 encode_sv (&enc, scalar); 894 encode_sv (&enc, scalar);
895 encode_nl (&enc);
647 896
648 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 897 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
649 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 898 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
650 899
651 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 900 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
669 JSON json; 918 JSON json;
670 U32 depth; // recursion depth 919 U32 depth; // recursion depth
671 U32 maxdepth; // recursion depth limit 920 U32 maxdepth; // recursion depth limit
672} dec_t; 921} dec_t;
673 922
674inline void 923INLINE void
924decode_comment (dec_t *dec)
925{
926 // only '#'-style comments allowed a.t.m.
927
928 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
929 ++dec->cur;
930}
931
932INLINE void
675decode_ws (dec_t *dec) 933decode_ws (dec_t *dec)
676{ 934{
677 for (;;) 935 for (;;)
678 { 936 {
679 char ch = *dec->cur; 937 char ch = *dec->cur;
680 938
681 if (ch > 0x20 939 if (ch > 0x20)
940 {
941 if (expect_false (ch == '#'))
942 {
943 if (dec->json.flags & F_RELAXED)
944 decode_comment (dec);
945 else
946 break;
947 }
948 else
949 break;
950 }
682 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 951 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
683 break; 952 break; // parse error, but let higher level handle it, gives better error messages
684 953
685 ++dec->cur; 954 ++dec->cur;
686 } 955 }
687} 956}
688 957
… …
692 if (*dec->cur != ch) \ 961 if (*dec->cur != ch) \
693 ERR (# ch " expected"); \ 962 ERR (# ch " expected"); \
694 ++dec->cur; \ 963 ++dec->cur; \
695 SE 964 SE
696 965
697#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 966#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
698#define DEC_DEC_DEPTH --dec->depth 967#define DEC_DEC_DEPTH --dec->depth
699 968
700static SV *decode_sv (dec_t *dec); 969static SV *decode_sv (dec_t *dec);
701 970
702static signed char decode_hexdigit[256]; 971static signed char decode_hexdigit[256];
… …
794 1063
795 if (hi >= 0x80) 1064 if (hi >= 0x80)
796 { 1065 {
797 utf8 = 1; 1066 utf8 = 1;
798 1067
799 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1068 cur = encode_utf8 (cur, hi);
800 } 1069 }
801 else 1070 else
802 *cur++ = hi; 1071 *cur++ = hi;
803 } 1072 }
804 break; 1073 break;
… …
806 default: 1075 default:
807 --dec_cur; 1076 --dec_cur;
808 ERR ("illegal backslash escape sequence in string"); 1077 ERR ("illegal backslash escape sequence in string");
809 } 1078 }
810 } 1079 }
811 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1080 else if (expect_true (ch >= 0x20 && ch < 0x80))
812 *cur++ = ch; 1081 *cur++ = ch;
813 else if (ch >= 0x80) 1082 else if (ch >= 0x80)
814 { 1083 {
815 STRLEN clen; 1084 STRLEN clen;
816 UV uch;
817 1085
818 --dec_cur; 1086 --dec_cur;
819 1087
820 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1088 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
821 if (clen == (STRLEN)-1) 1089 if (clen == (STRLEN)-1)
822 ERR ("malformed UTF-8 character in JSON string"); 1090 ERR ("malformed UTF-8 character in JSON string");
823 1091
824 do 1092 do
825 *cur++ = *dec_cur++; 1093 *cur++ = *dec_cur++;
… …
842 { 1110 {
843 STRLEN len = cur - buf; 1111 STRLEN len = cur - buf;
844 1112
845 if (sv) 1113 if (sv)
846 { 1114 {
847 SvGROW (sv, SvCUR (sv) + len + 1); 1115 STRLEN cur = SvCUR (sv);
1116
1117 if (SvLEN (sv) <= cur + len)
1118 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1119
848 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1120 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
849 SvCUR_set (sv, SvCUR (sv) + len); 1121 SvCUR_set (sv, SvCUR (sv) + len);
850 } 1122 }
851 else 1123 else
852 sv = newSVpvn (buf, len); 1124 sv = newSVpvn (buf, len);
… …
937 is_nv = 1; 1209 is_nv = 1;
938 } 1210 }
939 1211
940 if (!is_nv) 1212 if (!is_nv)
941 { 1213 {
1214 int len = dec->cur - start;
1215
942 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1216 // special case the rather common 1..5-digit-int case
943 if (*start == '-') 1217 if (*start == '-')
944 switch (dec->cur - start) 1218 switch (len)
945 { 1219 {
946 case 2: return newSViv (-( start [1] - '0' * 1)); 1220 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
947 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1221 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
948 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1222 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
949 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1223 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1224 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
950 } 1225 }
951 else 1226 else
952 switch (dec->cur - start) 1227 switch (len)
953 { 1228 {
954 case 1: return newSViv ( start [0] - '0' * 1); 1229 case 1: return newSViv ( start [0] - '0' * 1);
955 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1230 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
956 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1231 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
957 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1232 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1233 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
958 } 1234 }
959 1235
960 { 1236 {
961 UV uv; 1237 UV uv;
962 int numtype = grok_number (start, dec->cur - start, &uv); 1238 int numtype = grok_number (start, len, &uv);
963 if (numtype & IS_NUMBER_IN_UV) 1239 if (numtype & IS_NUMBER_IN_UV)
964 if (numtype & IS_NUMBER_NEG) 1240 if (numtype & IS_NUMBER_NEG)
965 { 1241 {
966 if (uv < (UV)IV_MIN) 1242 if (uv < (UV)IV_MIN)
967 return newSViv (-(IV)uv); 1243 return newSViv (-(IV)uv);
968 } 1244 }
969 else 1245 else
970 return newSVuv (uv); 1246 return newSVuv (uv);
971
972 // here would likely be the place for bigint support
973 } 1247 }
974 }
975 1248
976 // if we ever support bigint or bigfloat, this is the place for bigfloat 1249 len -= *start == '-' ? 1 : 0;
1250
1251 // does not fit into IV or UV, try NV
1252 if (len <= NV_DIG)
1253 // fits into NV without loss of precision
1254 return newSVnv (json_atof (start));
1255
1256 // everything else fails, convert it to a string
1257 return newSVpvn (start, dec->cur - start);
1258 }
1259
1260 // loss of precision here
977 return newSVnv (Atof (start)); 1261 return newSVnv (json_atof (start));
978 1262
979fail: 1263fail:
980 return 0; 1264 return 0;
981} 1265}
982 1266
… …
1006 if (*dec->cur == ']') 1290 if (*dec->cur == ']')
1007 { 1291 {
1008 ++dec->cur; 1292 ++dec->cur;
1009 break; 1293 break;
1010 } 1294 }
1011 1295
1012 if (*dec->cur != ',') 1296 if (*dec->cur != ',')
1013 ERR (", or ] expected while parsing array"); 1297 ERR (", or ] expected while parsing array");
1014 1298
1015 ++dec->cur; 1299 ++dec->cur;
1300
1301 decode_ws (dec);
1302
1303 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1304 {
1305 ++dec->cur;
1306 break;
1307 }
1016 } 1308 }
1017 1309
1018 DEC_DEC_DEPTH; 1310 DEC_DEC_DEPTH;
1019 return newRV_noinc ((SV *)av); 1311 return newRV_noinc ((SV *)av);
1020 1312
… …
1036 if (*dec->cur == '}') 1328 if (*dec->cur == '}')
1037 ++dec->cur; 1329 ++dec->cur;
1038 else 1330 else
1039 for (;;) 1331 for (;;)
1040 { 1332 {
1041 decode_ws (dec); EXPECT_CH ('"'); 1333 EXPECT_CH ('"');
1042 1334
1043 // heuristic: assume that 1335 // heuristic: assume that
1044 // a) decode_str + hv_store_ent are abysmally slow. 1336 // a) decode_str + hv_store_ent are abysmally slow.
1045 // b) most hash keys are short, simple ascii text. 1337 // b) most hash keys are short, simple ascii text.
1046 // => try to "fast-match" such strings to avoid 1338 // => try to "fast-match" such strings to avoid
… …
1050 char *p = dec->cur; 1342 char *p = dec->cur;
1051 char *e = p + 24; // only try up to 24 bytes 1343 char *e = p + 24; // only try up to 24 bytes
1052 1344
1053 for (;;) 1345 for (;;)
1054 { 1346 {
1055 // the >= 0x80 is true on most architectures 1347 // the >= 0x80 is false on most architectures
1056 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1348 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1057 { 1349 {
1058 // slow path, back up and use decode_str 1350 // slow path, back up and use decode_str
1059 SV *key = decode_str (dec); 1351 SV *key = decode_str (dec);
1060 if (!key) 1352 if (!key)
1061 goto fail; 1353 goto fail;
1062 1354
1063 decode_ws (dec); EXPECT_CH (':'); 1355 decode_ws (dec); EXPECT_CH (':');
1064 1356
1357 decode_ws (dec);
1065 value = decode_sv (dec); 1358 value = decode_sv (dec);
1066 if (!value) 1359 if (!value)
1067 { 1360 {
1068 SvREFCNT_dec (key); 1361 SvREFCNT_dec (key);
1069 goto fail; 1362 goto fail;
… …
1081 int len = p - key; 1374 int len = p - key;
1082 dec->cur = p + 1; 1375 dec->cur = p + 1;
1083 1376
1084 decode_ws (dec); EXPECT_CH (':'); 1377 decode_ws (dec); EXPECT_CH (':');
1085 1378
1379 decode_ws (dec);
1086 value = decode_sv (dec); 1380 value = decode_sv (dec);
1087 if (!value) 1381 if (!value)
1088 goto fail; 1382 goto fail;
1089 1383
1090 hv_store (hv, key, len, value, 0); 1384 hv_store (hv, key, len, value, 0);
… …
1106 1400
1107 if (*dec->cur != ',') 1401 if (*dec->cur != ',')
1108 ERR (", or } expected while parsing object/hash"); 1402 ERR (", or } expected while parsing object/hash");
1109 1403
1110 ++dec->cur; 1404 ++dec->cur;
1405
1406 decode_ws (dec);
1407
1408 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1409 {
1410 ++dec->cur;
1411 break;
1412 }
1111 } 1413 }
1112 1414
1113 DEC_DEC_DEPTH; 1415 DEC_DEC_DEPTH;
1114 sv = newRV_noinc ((SV *)hv); 1416 sv = newRV_noinc ((SV *)hv);
1115 1417
… …
1133 dSP; 1435 dSP;
1134 int count; 1436 int count;
1135 1437
1136 ENTER; SAVETMPS; PUSHMARK (SP); 1438 ENTER; SAVETMPS; PUSHMARK (SP);
1137 XPUSHs (HeVAL (he)); 1439 XPUSHs (HeVAL (he));
1440 sv_2mortal (sv);
1138 1441
1139 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1442 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1140 1443
1141 if (count == 1) 1444 if (count == 1)
1142 { 1445 {
1143 sv = newSVsv (POPs); 1446 sv = newSVsv (POPs);
1144 FREETMPS; LEAVE; 1447 FREETMPS; LEAVE;
1145 return sv; 1448 return sv;
1146 } 1449 }
1147 1450
1451 SvREFCNT_inc (sv);
1148 FREETMPS; LEAVE; 1452 FREETMPS; LEAVE;
1149 } 1453 }
1150 } 1454 }
1151 1455
1152 if (dec->json.cb_object) 1456 if (dec->json.cb_object)
… …
1178 DEC_DEC_DEPTH; 1482 DEC_DEC_DEPTH;
1179 return 0; 1483 return 0;
1180} 1484}
1181 1485
1182static SV * 1486static SV *
1487decode_tag (dec_t *dec)
1488{
1489 SV *tag = 0;
1490 SV *val = 0;
1491
1492 if (!(dec->json.flags & F_ALLOW_TAGS))
1493 ERR ("malformed JSON string, neither array, object, number, string or atom");
1494
1495 ++dec->cur;
1496
1497 decode_ws (dec);
1498
1499 tag = decode_sv (dec);
1500 if (!tag)
1501 goto fail;
1502
1503 if (!SvPOK (tag))
1504 ERR ("malformed JSON string, (tag) must be a string");
1505
1506 decode_ws (dec);
1507
1508 if (*dec->cur != ')')
1509 ERR (") expected after tag");
1510
1511 ++dec->cur;
1512
1513 decode_ws (dec);
1514
1515 val = decode_sv (dec);
1516 if (!val)
1517 goto fail;
1518
1519 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1520 ERR ("malformed JSON string, tag value must be an array");
1521
1522 {
1523 AV *av = (AV *)SvRV (val);
1524 int i, len = av_len (av) + 1;
1525 HV *stash = gv_stashsv (tag, 0);
1526 SV *sv;
1527
1528 if (!stash)
1529 ERR ("cannot decode perl-object (package does not exist)");
1530
1531 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1532
1533 if (!method)
1534 ERR ("cannot decode perl-object (package does not have a THAW method)");
1535
1536 dSP;
1537
1538 ENTER; SAVETMPS; PUSHMARK (SP);
1539 EXTEND (SP, len + 2);
1540 // we re-bless the reference to get overload and other niceties right
1541 PUSHs (tag);
1542 PUSHs (sv_json);
1543
1544 for (i = 0; i < len; ++i)
1545 PUSHs (*av_fetch (av, i, 1));
1546
1547 PUTBACK;
1548 call_sv ((SV *)GvCV (method), G_SCALAR);
1549 SPAGAIN;
1550
1551 SvREFCNT_dec (tag);
1552 SvREFCNT_dec (val);
1553 sv = SvREFCNT_inc (POPs);
1554
1555 PUTBACK;
1556
1557 FREETMPS; LEAVE;
1558
1559 return sv;
1560 }
1561
1562fail:
1563 SvREFCNT_dec (tag);
1564 SvREFCNT_dec (val);
1565 return 0;
1566}
1567
1568static SV *
1183decode_sv (dec_t *dec) 1569decode_sv (dec_t *dec)
1184{ 1570{
1185 decode_ws (dec);
1186
1187 // the beauty of JSON: you need exactly one character lookahead 1571 // the beauty of JSON: you need exactly one character lookahead
1188 // to parse anything. 1572 // to parse everything.
1189 switch (*dec->cur) 1573 switch (*dec->cur)
1190 { 1574 {
1191 case '"': ++dec->cur; return decode_str (dec); 1575 case '"': ++dec->cur; return decode_str (dec);
1192 case '[': ++dec->cur; return decode_av (dec); 1576 case '[': ++dec->cur; return decode_av (dec);
1193 case '{': ++dec->cur; return decode_hv (dec); 1577 case '{': ++dec->cur; return decode_hv (dec);
1578 case '(': return decode_tag (dec);
1194 1579
1195 case '-': 1580 case '-':
1196 case '0': case '1': case '2': case '3': case '4': 1581 case '0': case '1': case '2': case '3': case '4':
1197 case '5': case '6': case '7': case '8': case '9': 1582 case '5': case '6': case '7': case '8': case '9':
1198 return decode_num (dec); 1583 return decode_num (dec);
1199 1584
1200 case 't': 1585 case 't':
1201 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1586 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1202 { 1587 {
1203 dec->cur += 4; 1588 dec->cur += 4;
1204 return SvREFCNT_inc (json_true); 1589#if JSON_SLOW
1590 types_true = get_bool ("Types::Serialiser::true");
1591#endif
1592 return newSVsv (types_true);
1205 } 1593 }
1206 else 1594 else
1207 ERR ("'true' expected"); 1595 ERR ("'true' expected");
1208 1596
1209 break; 1597 break;
1210 1598
1211 case 'f': 1599 case 'f':
1212 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1600 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1213 { 1601 {
1214 dec->cur += 5; 1602 dec->cur += 5;
1215 return SvREFCNT_inc (json_false); 1603#if JSON_SLOW
1604 types_false = get_bool ("Types::Serialiser::false");
1605#endif
1606 return newSVsv (types_false);
1216 } 1607 }
1217 else 1608 else
1218 ERR ("'false' expected"); 1609 ERR ("'false' expected");
1219 1610
1220 break; 1611 break;
… …
1229 ERR ("'null' expected"); 1620 ERR ("'null' expected");
1230 1621
1231 break; 1622 break;
1232 1623
1233 default: 1624 default:
1234 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1625 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1235 break; 1626 break;
1236 } 1627 }
1237 1628
1238fail: 1629fail:
1239 return 0; 1630 return 0;
1240} 1631}
1241 1632
1242static SV * 1633static SV *
1243decode_json (SV *string, JSON *json, UV *offset_return) 1634decode_json (SV *string, JSON *json, char **offset_return)
1244{ 1635{
1245 dec_t dec; 1636 dec_t dec;
1246 UV offset;
1247 SV *sv; 1637 SV *sv;
1248 1638
1639 /* work around bugs in 5.10 where manipulating magic values
1640 * makes perl ignore the magic in subsequent accesses.
1641 * also make a copy of non-PV values, to get them into a clean
1642 * state (SvPV should do that, but it's buggy, see below).
1643 */
1249 SvGETMAGIC (string); 1644 /*SvGETMAGIC (string);*/
1645 if (SvMAGICAL (string) || !SvPOK (string))
1646 string = sv_2mortal (newSVsv (string));
1647
1250 SvUPGRADE (string, SVt_PV); 1648 SvUPGRADE (string, SVt_PV);
1251 1649
1252 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1650 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1651 * assertion with -DDEBUGGING, although SvCUR is documented to
1652 * return the xpv_cur field which certainly exists after upgrading.
1653 * according to nicholas clark, calling SvPOK fixes this.
1654 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1655 * and hope for the best.
1656 * Damnit, SvPV_nolen still trips over yet another assertion. This
1657 * assertion business is seriously broken, try yet another workaround
1658 * for the broken -DDEBUGGING.
1659 */
1660 {
1661#ifdef DEBUGGING
1662 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1663#else
1664 STRLEN offset = SvCUR (string);
1665#endif
1666
1667 if (offset > json->max_size && json->max_size)
1253 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1668 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1254 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1669 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1670 }
1255 1671
1256 if (json->flags & F_UTF8) 1672 if (DECODE_WANTS_OCTETS (json))
1257 sv_utf8_downgrade (string, 0); 1673 sv_utf8_downgrade (string, 0);
1258 else 1674 else
1259 sv_utf8_upgrade (string); 1675 sv_utf8_upgrade (string);
1260 1676
1261 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1677 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1262 1678
1263 dec.json = *json; 1679 dec.json = *json;
1264 dec.cur = SvPVX (string); 1680 dec.cur = SvPVX (string);
1265 dec.end = SvEND (string); 1681 dec.end = SvEND (string);
1266 dec.err = 0; 1682 dec.err = 0;
1267 dec.depth = 0; 1683 dec.depth = 0;
1268 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1269 1684
1270 if (dec.json.cb_object || dec.json.cb_sk_object) 1685 if (dec.json.cb_object || dec.json.cb_sk_object)
1271 dec.json.flags |= F_HOOK; 1686 dec.json.flags |= F_HOOK;
1272 1687
1273 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1688 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1689
1690 decode_ws (&dec);
1274 sv = decode_sv (&dec); 1691 sv = decode_sv (&dec);
1692
1693 if (offset_return)
1694 *offset_return = dec.cur;
1275 1695
1276 if (!(offset_return || !sv)) 1696 if (!(offset_return || !sv))
1277 { 1697 {
1278 // check for trailing garbage 1698 // check for trailing garbage
1279 decode_ws (&dec); 1699 decode_ws (&dec);
… …
1282 { 1702 {
1283 dec.err = "garbage after JSON object"; 1703 dec.err = "garbage after JSON object";
1284 SvREFCNT_dec (sv); 1704 SvREFCNT_dec (sv);
1285 sv = 0; 1705 sv = 0;
1286 } 1706 }
1287 }
1288
1289 if (offset_return || !sv)
1290 {
1291 offset = dec.json.flags & F_UTF8
1292 ? dec.cur - SvPVX (string)
1293 : utf8_distance (dec.cur, SvPVX (string));
1294
1295 if (offset_return)
1296 *offset_return = offset;
1297 } 1707 }
1298 1708
1299 if (!sv) 1709 if (!sv)
1300 { 1710 {
1301 SV *uni = sv_newmortal (); 1711 SV *uni = sv_newmortal ();
… …
1307 SAVEVPTR (PL_curcop); 1717 SAVEVPTR (PL_curcop);
1308 PL_curcop = &cop; 1718 PL_curcop = &cop;
1309 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1719 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1310 LEAVE; 1720 LEAVE;
1311 1721
1312 croak ("%s, at character offset %d [\"%s\"]", 1722 croak ("%s, at character offset %d (before \"%s\")",
1313 dec.err, 1723 dec.err,
1314 (int)offset, 1724 (int)ptr_to_index (string, dec.cur),
1315 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1725 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1316 } 1726 }
1317 1727
1318 sv = sv_2mortal (sv); 1728 sv = sv_2mortal (sv);
1319 1729
1320 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1730 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1321 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1731 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1322 1732
1323 return sv; 1733 return sv;
1734}
1735
1736/////////////////////////////////////////////////////////////////////////////
1737// incremental parser
1738
1739static void
1740incr_parse (JSON *self)
1741{
1742 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1743
1744 // the state machine here is a bit convoluted and could be simplified a lot
1745 // but this would make it slower, so...
1746
1747 for (;;)
1748 {
1749 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D
1750 switch (self->incr_mode)
1751 {
1752 // only used for initial whitespace skipping
1753 case INCR_M_WS:
1754 for (;;)
1755 {
1756 if (*p > 0x20)
1757 {
1758 if (*p == '#')
1759 {
1760 self->incr_mode = INCR_M_C0;
1761 goto incr_m_c;
1762 }
1763 else
1764 {
1765 self->incr_mode = INCR_M_JSON;
1766 goto incr_m_json;
1767 }
1768 }
1769 else if (!*p)
1770 goto interrupt;
1771
1772 ++p;
1773 }
1774
1775 // skip a single char inside a string (for \\-processing)
1776 case INCR_M_BS:
1777 if (!*p)
1778 goto interrupt;
1779
1780 ++p;
1781 self->incr_mode = INCR_M_STR;
1782 goto incr_m_str;
1783
1784 // inside #-style comments
1785 case INCR_M_C0:
1786 case INCR_M_C1:
1787 incr_m_c:
1788 for (;;)
1789 {
1790 if (*p == '\n')
1791 {
1792 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1793 break;
1794 }
1795 else if (!*p)
1796 goto interrupt;
1797
1798 ++p;
1799 }
1800
1801 break;
1802
1803 // inside a string
1804 case INCR_M_STR:
1805 incr_m_str:
1806 for (;;)
1807 {
1808 if (*p == '"')
1809 {
1810 ++p;
1811 self->incr_mode = INCR_M_JSON;
1812
1813 if (!self->incr_nest)
1814 goto interrupt;
1815
1816 goto incr_m_json;
1817 }
1818 else if (*p == '\\')
1819 {
1820 ++p; // "virtually" consumes character after \
1821
1822 if (!*p) // if at end of string we have to switch modes
1823 {
1824 self->incr_mode = INCR_M_BS;
1825 goto interrupt;
1826 }
1827 }
1828 else if (!*p)
1829 goto interrupt;
1830
1831 ++p;
1832 }
1833
1834 // after initial ws, outside string
1835 case INCR_M_JSON:
1836 incr_m_json:
1837 for (;;)
1838 {
1839 switch (*p++)
1840 {
1841 case 0:
1842 --p;
1843 goto interrupt;
1844
1845 case 0x09:
1846 case 0x0a:
1847 case 0x0d:
1848 case 0x20:
1849 if (!self->incr_nest)
1850 {
1851 --p; // do not eat the whitespace, let the next round do it
1852 goto interrupt;
1853 }
1854 break;
1855
1856 case '"':
1857 self->incr_mode = INCR_M_STR;
1858 goto incr_m_str;
1859
1860 case '[':
1861 case '{':
1862 case '(':
1863 if (++self->incr_nest > self->max_depth)
1864 croak (ERR_NESTING_EXCEEDED);
1865 break;
1866
1867 case ']':
1868 case '}':
1869 if (--self->incr_nest <= 0)
1870 goto interrupt;
1871 break;
1872
1873 case ')':
1874 --self->incr_nest;
1875 break;
1876
1877 case '#':
1878 self->incr_mode = INCR_M_C1;
1879 goto incr_m_c;
1880 }
1881 }
1882 }
1883
1884 modechange:
1885 ;
1886 }
1887
1888interrupt:
1889 self->incr_pos = p - SvPVX (self->incr_text);
1890 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1891 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1324} 1892}
1325 1893
1326///////////////////////////////////////////////////////////////////////////// 1894/////////////////////////////////////////////////////////////////////////////
1327// XS interface functions 1895// XS interface functions
1328 1896
… …
1337 i >= '0' && i <= '9' ? i - '0' 1905 i >= '0' && i <= '9' ? i - '0'
1338 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1906 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1339 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1907 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1340 : -1; 1908 : -1;
1341 1909
1342 json_stash = gv_stashpv ("JSON::XS" , 1); 1910 json_stash = gv_stashpv ("JSON::XS" , 1);
1343 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1911 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1344 1912
1345 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1913 types_true = get_bool ("Types::Serialiser::true");
1346 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1914 types_false = get_bool ("Types::Serialiser::false");
1915
1916 sv_json = newSVpv ("JSON", 0);
1917 SvREADONLY_on (sv_json);
1918
1919 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1347} 1920}
1348 1921
1349PROTOTYPES: DISABLE 1922PROTOTYPES: DISABLE
1923
1924void CLONE (...)
1925 CODE:
1926 json_stash = 0;
1927 types_boolean_stash = 0;
1350 1928
1351void new (char *klass) 1929void new (char *klass)
1352 PPCODE: 1930 PPCODE:
1353{ 1931{
1354 SV *pv = NEWSV (0, sizeof (JSON)); 1932 SV *pv = NEWSV (0, sizeof (JSON));
1355 SvPOK_only (pv); 1933 SvPOK_only (pv);
1356 Zero (SvPVX (pv), 1, JSON); 1934 json_init ((JSON *)SvPVX (pv));
1357 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1935 XPUSHs (sv_2mortal (sv_bless (
1358 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash))); 1936 newRV_noinc (pv),
1937 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1938 )));
1359} 1939}
1360 1940
1361void ascii (JSON *self, int enable = 1) 1941void ascii (JSON *self, int enable = 1)
1362 ALIAS: 1942 ALIAS:
1363 ascii = F_ASCII 1943 ascii = F_ASCII
… …
1370 pretty = F_PRETTY 1950 pretty = F_PRETTY
1371 allow_nonref = F_ALLOW_NONREF 1951 allow_nonref = F_ALLOW_NONREF
1372 shrink = F_SHRINK 1952 shrink = F_SHRINK
1373 allow_blessed = F_ALLOW_BLESSED 1953 allow_blessed = F_ALLOW_BLESSED
1374 convert_blessed = F_CONV_BLESSED 1954 convert_blessed = F_CONV_BLESSED
1955 relaxed = F_RELAXED
1956 allow_unknown = F_ALLOW_UNKNOWN
1957 allow_tags = F_ALLOW_TAGS
1375 PPCODE: 1958 PPCODE:
1376{ 1959{
1377 if (enable) 1960 if (enable)
1378 self->flags |= ix; 1961 self->flags |= ix;
1379 else 1962 else
1380 self->flags &= ~ix; 1963 self->flags &= ~ix;
1381 1964
1382 XPUSHs (ST (0)); 1965 XPUSHs (ST (0));
1383} 1966}
1384 1967
1385void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1968void get_ascii (JSON *self)
1969 ALIAS:
1970 get_ascii = F_ASCII
1971 get_latin1 = F_LATIN1
1972 get_utf8 = F_UTF8
1973 get_indent = F_INDENT
1974 get_canonical = F_CANONICAL
1975 get_space_before = F_SPACE_BEFORE
1976 get_space_after = F_SPACE_AFTER
1977 get_allow_nonref = F_ALLOW_NONREF
1978 get_shrink = F_SHRINK
1979 get_allow_blessed = F_ALLOW_BLESSED
1980 get_convert_blessed = F_CONV_BLESSED
1981 get_relaxed = F_RELAXED
1982 get_allow_unknown = F_ALLOW_UNKNOWN
1983 get_allow_tags = F_ALLOW_TAGS
1386 PPCODE: 1984 PPCODE:
1387{ 1985 XPUSHs (boolSV (self->flags & ix));
1388 UV log2 = 0;
1389 1986
1390 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1987void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1391 1988 PPCODE:
1392 while ((1UL << log2) < max_depth) 1989 self->max_depth = max_depth;
1393 ++log2;
1394
1395 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1396
1397 XPUSHs (ST (0)); 1990 XPUSHs (ST (0));
1398}
1399 1991
1992U32 get_max_depth (JSON *self)
1993 CODE:
1994 RETVAL = self->max_depth;
1995 OUTPUT:
1996 RETVAL
1997
1400void max_size (JSON *self, UV max_size = 0) 1998void max_size (JSON *self, U32 max_size = 0)
1401 PPCODE: 1999 PPCODE:
1402{ 2000 self->max_size = max_size;
1403 UV log2 = 0;
1404
1405 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1406 if (max_size == 1) max_size = 2;
1407
1408 while ((1UL << log2) < max_size)
1409 ++log2;
1410
1411 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1412
1413 XPUSHs (ST (0)); 2001 XPUSHs (ST (0));
1414} 2002
2003int get_max_size (JSON *self)
2004 CODE:
2005 RETVAL = self->max_size;
2006 OUTPUT:
2007 RETVAL
1415 2008
1416void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 2009void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1417 PPCODE: 2010 PPCODE:
1418{ 2011{
1419 SvREFCNT_dec (self->cb_object); 2012 SvREFCNT_dec (self->cb_object);
… …
1423} 2016}
1424 2017
1425void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2018void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1426 PPCODE: 2019 PPCODE:
1427{ 2020{
1428 if (!self->cb_sk_object) 2021 if (!self->cb_sk_object)
1429 self->cb_sk_object = newHV (); 2022 self->cb_sk_object = newHV ();
1430 2023
1431 if (SvOK (cb)) 2024 if (SvOK (cb))
1432 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2025 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1433 else 2026 else
… …
1444 XPUSHs (ST (0)); 2037 XPUSHs (ST (0));
1445} 2038}
1446 2039
1447void encode (JSON *self, SV *scalar) 2040void encode (JSON *self, SV *scalar)
1448 PPCODE: 2041 PPCODE:
1449 XPUSHs (encode_json (scalar, self)); 2042 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2043 XPUSHs (scalar);
1450 2044
1451void decode (JSON *self, SV *jsonstr) 2045void decode (JSON *self, SV *jsonstr)
1452 PPCODE: 2046 PPCODE:
1453 XPUSHs (decode_json (jsonstr, self, 0)); 2047 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2048 XPUSHs (jsonstr);
1454 2049
1455void decode_prefix (JSON *self, SV *jsonstr) 2050void decode_prefix (JSON *self, SV *jsonstr)
1456 PPCODE: 2051 PPCODE:
1457{ 2052{
1458 UV offset; 2053 SV *sv;
2054 char *offset;
2055 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1459 EXTEND (SP, 2); 2056 EXTEND (SP, 2);
1460 PUSHs (decode_json (jsonstr, self, &offset)); 2057 PUSHs (sv);
1461 PUSHs (sv_2mortal (newSVuv (offset))); 2058 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
2059}
2060
2061void incr_parse (JSON *self, SV *jsonstr = 0)
2062 PPCODE:
2063{
2064 if (!self->incr_text)
2065 self->incr_text = newSVpvn ("", 0);
2066
2067 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2068 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2069 if (DECODE_WANTS_OCTETS (self))
2070 {
2071 if (self->incr_pos)
2072 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2073 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2074
2075 sv_utf8_downgrade (self->incr_text, 0);
2076 }
2077 else
2078 {
2079 sv_utf8_upgrade (self->incr_text);
2080
2081 if (self->incr_pos)
2082 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2083 - (U8 *)SvPVX (self->incr_text);
2084 }
2085
2086 // append data, if any
2087 if (jsonstr)
2088 {
2089 /* make sure both strings have same encoding */
2090 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2091 if (SvUTF8 (jsonstr))
2092 sv_utf8_downgrade (jsonstr, 0);
2093 else
2094 sv_utf8_upgrade (jsonstr);
2095
2096 /* and then just blindly append */
2097 {
2098 STRLEN len;
2099 const char *str = SvPV (jsonstr, len);
2100 STRLEN cur = SvCUR (self->incr_text);
2101
2102 if (SvLEN (self->incr_text) <= cur + len)
2103 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
2104
2105 Move (str, SvEND (self->incr_text), len, char);
2106 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2107 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2108 }
2109 }
2110
2111 if (GIMME_V != G_VOID)
2112 do
2113 {
2114 SV *sv;
2115 char *offset;
2116
2117 if (!INCR_DONE (self))
2118 {
2119 incr_parse (self);
2120
2121 if (self->incr_pos > self->max_size && self->max_size)
2122 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2123 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2124
2125 if (!INCR_DONE (self))
2126 {
2127 // as an optimisation, do not accumulate white space in the incr buffer
2128 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2129 {
2130 self->incr_pos = 0;
2131 SvCUR_set (self->incr_text, 0);
2132 }
2133
2134 break;
2135 }
2136 }
2137
2138 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2139 XPUSHs (sv);
2140
2141 self->incr_pos -= offset - SvPVX (self->incr_text);
2142 self->incr_nest = 0;
2143 self->incr_mode = 0;
2144
2145 sv_chop (self->incr_text, offset);
2146 }
2147 while (GIMME_V == G_ARRAY);
2148}
2149
2150SV *incr_text (JSON *self)
2151 ATTRS: lvalue
2152 CODE:
2153{
2154 if (self->incr_pos)
2155 croak ("incr_text can not be called when the incremental parser already started parsing");
2156
2157 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2158}
2159 OUTPUT:
2160 RETVAL
2161
2162void incr_skip (JSON *self)
2163 CODE:
2164{
2165 if (self->incr_pos)
2166 {
2167 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2168 self->incr_pos = 0;
2169 self->incr_nest = 0;
2170 self->incr_mode = 0;
2171 }
2172}
2173
2174void incr_reset (JSON *self)
2175 CODE:
2176{
2177 SvREFCNT_dec (self->incr_text);
2178 self->incr_text = 0;
2179 self->incr_pos = 0;
2180 self->incr_nest = 0;
2181 self->incr_mode = 0;
1462} 2182}
1463 2183
1464void DESTROY (JSON *self) 2184void DESTROY (JSON *self)
1465 CODE: 2185 CODE:
1466 SvREFCNT_dec (self->cb_sk_object); 2186 SvREFCNT_dec (self->cb_sk_object);
1467 SvREFCNT_dec (self->cb_object); 2187 SvREFCNT_dec (self->cb_object);
2188 SvREFCNT_dec (self->incr_text);
1468 2189
1469PROTOTYPES: ENABLE 2190PROTOTYPES: ENABLE
1470 2191
1471void to_json (SV *scalar) 2192void encode_json (SV *scalar)
1472 PPCODE: 2193 PPCODE:
1473{ 2194{
1474 JSON json = { F_DEFAULT | F_UTF8 }; 2195 JSON json;
1475 XPUSHs (encode_json (scalar, &json)); 2196 json_init (&json);
2197 json.flags |= F_UTF8;
2198 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2199 XPUSHs (scalar);
1476} 2200}
1477 2201
1478void from_json (SV *jsonstr) 2202void decode_json (SV *jsonstr)
1479 PPCODE: 2203 PPCODE:
1480{ 2204{
1481 JSON json = { F_DEFAULT | F_UTF8 }; 2205 JSON json;
1482 XPUSHs (decode_json (jsonstr, &json, 0)); 2206 json_init (&json);
2207 json.flags |= F_UTF8;
2208 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2209 XPUSHs (jsonstr);
1483} 2210}
1484 2211

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines