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 1.63 by root, Mon Aug 27 01:49:01 2007 UTC vs.
Revision 1.129 by root, Wed Nov 16 18:06:34 2016 UTC

… …
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>
9#include <float.h> 10#include <float.h>
10 11
11#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
13#endif 14#endif
14 15
15// 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
16// guarentees, though. if it breaks, you get to keep the pieces. 17// guarantees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
19#endif 20#endif
21
22// compatibility with perl <5.18
23#ifndef HvNAMELEN_get
24# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
25#endif
26#ifndef HvNAMELEN
27# define HvNAMELEN(hv) HvNAMELEN_get (hv)
28#endif
29#ifndef HvNAMEUTF8
30# define HvNAMEUTF8(hv) 0
31#endif
32
33// three extra for rounding, sign, and end of string
34#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
20 35
21#define F_ASCII 0x00000001UL 36#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL 37#define F_LATIN1 0x00000002UL
23#define F_UTF8 0x00000004UL 38#define F_UTF8 0x00000004UL
24#define F_INDENT 0x00000008UL 39#define F_INDENT 0x00000008UL
… …
28#define F_ALLOW_NONREF 0x00000100UL 43#define F_ALLOW_NONREF 0x00000100UL
29#define F_SHRINK 0x00000200UL 44#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL 45#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL 46#define F_CONV_BLESSED 0x00000800UL
32#define F_RELAXED 0x00001000UL 47#define F_RELAXED 0x00001000UL
33 48#define F_ALLOW_UNKNOWN 0x00002000UL
34#define F_MAXDEPTH 0xf8000000UL 49#define F_ALLOW_TAGS 0x00004000UL
35#define S_MAXDEPTH 27
36#define F_MAXSIZE 0x01f00000UL
37#define S_MAXSIZE 20
38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 50#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
39 51
40#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
41#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
42
43#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 52#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
44#define F_DEFAULT (9UL << S_MAXDEPTH)
45 53
46#define INIT_SIZE 32 // initial scalar size to be allocated 54#define INIT_SIZE 32 // initial scalar size to be allocated
47#define INDENT_STEP 3 // spaces per indentation level 55#define INDENT_STEP 3 // spaces per indentation level
48 56
49#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 57#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
50 58
59#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
60
51#define SB do { 61#define SB do {
52#define SE } while (0) 62#define SE } while (0)
53 63
54#if __GNUC__ >= 3 64#if __GNUC__ >= 3
55# define expect(expr,value) __builtin_expect ((expr),(value)) 65# define expect(expr,value) __builtin_expect ((expr), (value))
56# define inline inline 66# define INLINE static inline
57#else 67#else
58# define expect(expr,value) (expr) 68# define expect(expr,value) (expr)
59# define inline static 69# define INLINE static
60#endif 70#endif
61 71
62#define expect_false(expr) expect ((expr) != 0, 0) 72#define expect_false(expr) expect ((expr) != 0, 0)
63#define expect_true(expr) expect ((expr) != 0, 1) 73#define expect_true(expr) expect ((expr) != 0, 1)
74
75#define IN_RANGE_INC(type,val,beg,end) \
76 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
77 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
78
79#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
64 80
65#ifdef USE_ITHREADS 81#ifdef USE_ITHREADS
66# define JSON_SLOW 1 82# define JSON_SLOW 1
67# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 83# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
84# define BOOL_STASH (bool_stash ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1))
68#else 85#else
69# define JSON_SLOW 0 86# define JSON_SLOW 0
70# define JSON_STASH json_stash 87# define JSON_STASH json_stash
88# define BOOL_STASH bool_stash
71#endif 89#endif
72 90
73static HV *json_stash, *json_boolean_stash; // JSON::XS:: 91// the amount of HEs to allocate on the stack, when sorting keys
74static SV *json_true, *json_false; 92#define STACK_HES 64
93
94static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
95static SV *bool_true, *bool_false, *sv_json;
96
97enum {
98 INCR_M_WS = 0, // initial whitespace skipping, must be 0
99 INCR_M_STR, // inside string
100 INCR_M_BS, // inside backslash
101 INCR_M_C0, // inside comment in initial whitespace sequence
102 INCR_M_C1, // inside comment in other places
103 INCR_M_JSON // outside anything, count nesting
104};
105
106#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
75 107
76typedef struct { 108typedef struct {
77 U32 flags; 109 U32 flags;
110 U32 max_depth;
111 STRLEN max_size;
112
78 SV *cb_object; 113 SV *cb_object;
79 HV *cb_sk_object; 114 HV *cb_sk_object;
115
116 // for the incremental parser
117 SV *incr_text; // the source text so far
118 STRLEN incr_pos; // the current offset into the text
119 int incr_nest; // {[]}-nesting level
120 unsigned char incr_mode;
80} JSON; 121} JSON;
122
123INLINE void
124json_init (JSON *json)
125{
126 Zero (json, 1, JSON);
127 json->max_depth = 512;
128}
81 129
82///////////////////////////////////////////////////////////////////////////// 130/////////////////////////////////////////////////////////////////////////////
83// utility functions 131// utility functions
84 132
85inline void 133INLINE SV *
134get_bool (const char *name)
135{
136 SV *sv = get_sv (name, 1);
137
138 SvREADONLY_on (sv);
139 SvREADONLY_on (SvRV (sv));
140
141 return sv;
142}
143
144INLINE void
86shrink (SV *sv) 145shrink (SV *sv)
87{ 146{
88 sv_utf8_downgrade (sv, 1); 147 sv_utf8_downgrade (sv, 1);
148
89 if (SvLEN (sv) > SvCUR (sv) + 1) 149 if (SvLEN (sv) > SvCUR (sv) + 1)
90 { 150 {
91#ifdef SvPV_shrink_to_cur 151#ifdef SvPV_shrink_to_cur
92 SvPV_shrink_to_cur (sv); 152 SvPV_shrink_to_cur (sv);
93#elif defined (SvPV_renew) 153#elif defined (SvPV_renew)
… …
99// decode an utf-8 character and return it, or (UV)-1 in 159// decode an utf-8 character and return it, or (UV)-1 in
100// case of an error. 160// case of an error.
101// we special-case "safe" characters from U+80 .. U+7FF, 161// we special-case "safe" characters from U+80 .. U+7FF,
102// but use the very good perl function to parse anything else. 162// but use the very good perl function to parse anything else.
103// note that we never call this function for a ascii codepoints 163// note that we never call this function for a ascii codepoints
104inline UV 164INLINE UV
105decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 165decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
106{ 166{
107 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 167 if (expect_true (len >= 2
108 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 168 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
109 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 169 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
110 { 170 {
111 *clen = 2; 171 *clen = 2;
112 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 172 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
113 } 173 }
114 else 174 else
175 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
176}
177
178// likewise for encoding, also never called for ascii codepoints
179// this function takes advantage of this fact, although current gccs
180// seem to optimise the check for >= 0x80 away anyways
181INLINE unsigned char *
182encode_utf8 (unsigned char *s, UV ch)
183{
184 if (expect_false (ch < 0x000080))
185 *s++ = ch;
186 else if (expect_true (ch < 0x000800))
187 *s++ = 0xc0 | ( ch >> 6),
188 *s++ = 0x80 | ( ch & 0x3f);
189 else if ( ch < 0x010000)
190 *s++ = 0xe0 | ( ch >> 12),
191 *s++ = 0x80 | ((ch >> 6) & 0x3f),
192 *s++ = 0x80 | ( ch & 0x3f);
193 else if ( ch < 0x110000)
194 *s++ = 0xf0 | ( ch >> 18),
195 *s++ = 0x80 | ((ch >> 12) & 0x3f),
196 *s++ = 0x80 | ((ch >> 6) & 0x3f),
197 *s++ = 0x80 | ( ch & 0x3f);
198
199 return s;
200}
201
202// convert offset pointer to character index, sv must be string
203static STRLEN
204ptr_to_index (SV *sv, char *offset)
205{
206 return SvUTF8 (sv)
207 ? utf8_distance (offset, SvPVX (sv))
208 : offset - SvPVX (sv);
209}
210
211/////////////////////////////////////////////////////////////////////////////
212// fp hell
213
214// scan a group of digits, and a trailing exponent
215static void
216json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
217{
218 UV uaccum = 0;
219 int eaccum = 0;
220
221 // if we recurse too deep, skip all remaining digits
222 // to avoid a stack overflow attack
223 if (expect_false (--maxdepth <= 0))
224 while (((U8)*s - '0') < 10)
225 ++s;
226
227 for (;;)
228 {
229 U8 dig = (U8)*s - '0';
230
231 if (expect_false (dig >= 10))
232 {
233 if (dig == (U8)((U8)'.' - (U8)'0'))
234 {
235 ++s;
236 json_atof_scan1 (s, accum, expo, 1, maxdepth);
237 }
238 else if ((dig | ' ') == 'e' - '0')
239 {
240 int exp2 = 0;
241 int neg = 0;
242
243 ++s;
244
245 if (*s == '-')
246 {
247 ++s;
248 neg = 1;
249 }
250 else if (*s == '+')
251 ++s;
252
253 while ((dig = (U8)*s - '0') < 10)
254 exp2 = exp2 * 10 + *s++ - '0';
255
256 *expo += neg ? -exp2 : exp2;
257 }
258
259 break;
260 }
261
262 ++s;
263
264 uaccum = uaccum * 10 + dig;
265 ++eaccum;
266
267 // if we have too many digits, then recurse for more
268 // we actually do this for rather few digits
269 if (uaccum >= (UV_MAX - 9) / 10)
270 {
271 if (postdp) *expo -= eaccum;
272 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
273 if (postdp) *expo += eaccum;
274
275 break;
276 }
115 { 277 }
116 *clen = (STRLEN)-1; 278
279 // this relies greatly on the quality of the pow ()
280 // implementation of the platform, but a good
281 // implementation is hard to beat.
282 // (IEEE 754 conformant ones are required to be exact)
283 if (postdp) *expo -= eaccum;
284 *accum += uaccum * Perl_pow (10., *expo);
285 *expo += eaccum;
286}
287
288static NV
289json_atof (const char *s)
290{
291 NV accum = 0.;
292 int expo = 0;
293 int neg = 0;
294
295 if (*s == '-')
296 {
297 ++s;
298 neg = 1;
299 }
300
301 // a recursion depth of ten gives us >>500 bits
302 json_atof_scan1 (s, &accum, &expo, 0, 10);
303
304 return neg ? -accum : accum;
305}
306
307// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
308static int
309ref_bool_type (SV *sv)
310{
311 svtype svt = SvTYPE (sv);
312
313 if (svt < SVt_PVAV)
314 {
315 STRLEN len = 0;
316 char *pv = svt ? SvPV (sv, len) : 0;
317
318 if (len == 1)
319 if (*pv == '1')
320 return 1;
321 else if (*pv == '0')
322 return 0;
323 }
324
325 return -1;
326}
327
328// returns whether scalar is not a reference in the sense of allow_nonref
329static int
330json_nonref (SV *scalar)
331{
332 if (!SvROK (scalar))
333 return 1;
334
335 scalar = SvRV (scalar);
336
337 if (SvTYPE (scalar) >= SVt_PVMG)
338 {
339 if (SvSTASH (scalar) == bool_stash)
117 return (UV)-1; 340 return 1;
341
342 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
343 return 1;
118 } 344 }
345
346 return 0;
119} 347}
120 348
121///////////////////////////////////////////////////////////////////////////// 349/////////////////////////////////////////////////////////////////////////////
122// encoder 350// encoder
123 351
… …
127 char *cur; // SvPVX (sv) + current output position 355 char *cur; // SvPVX (sv) + current output position
128 char *end; // SvEND (sv) 356 char *end; // SvEND (sv)
129 SV *sv; // result scalar 357 SV *sv; // result scalar
130 JSON json; 358 JSON json;
131 U32 indent; // indentation level 359 U32 indent; // indentation level
132 U32 maxdepth; // max. indentation/recursion level 360 UV limit; // escape character values >= this value when encoding
133} enc_t; 361} enc_t;
134 362
135inline void 363INLINE void
136need (enc_t *enc, STRLEN len) 364need (enc_t *enc, STRLEN len)
137{ 365{
138 if (expect_false (enc->cur + len >= enc->end)) 366 if (expect_false (enc->cur + len >= enc->end))
139 { 367 {
140 STRLEN cur = enc->cur - SvPVX (enc->sv); 368 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
141 SvGROW (enc->sv, cur + len + 1); 369 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
142 enc->cur = SvPVX (enc->sv) + cur; 370 enc->cur = SvPVX (enc->sv) + cur;
143 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 371 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
144 } 372 }
145} 373}
146 374
147inline void 375INLINE void
148encode_ch (enc_t *enc, char ch) 376encode_ch (enc_t *enc, char ch)
149{ 377{
150 need (enc, 1); 378 need (enc, 1);
151 *enc->cur++ = ch; 379 *enc->cur++ = ch;
152} 380}
… …
206 { 434 {
207 uch = ch; 435 uch = ch;
208 clen = 1; 436 clen = 1;
209 } 437 }
210 438
211 if (uch > 0x10FFFFUL) 439 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
212 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
213
214 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
215 { 440 {
216 if (uch > 0xFFFFUL) 441 if (uch >= 0x10000UL)
217 { 442 {
443 if (uch >= 0x110000UL)
444 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
445
218 need (enc, len += 11); 446 need (enc, len += 11);
219 sprintf (enc->cur, "\\u%04x\\u%04x", 447 sprintf (enc->cur, "\\u%04x\\u%04x",
220 (int)((uch - 0x10000) / 0x400 + 0xD800), 448 (int)((uch - 0x10000) / 0x400 + 0xD800),
221 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 449 (int)((uch - 0x10000) % 0x400 + 0xDC00));
222 enc->cur += 12; 450 enc->cur += 12;
223 } 451 }
224 else 452 else
225 { 453 {
226 static char hexdigit [16] = "0123456789abcdef";
227 need (enc, len += 5); 454 need (enc, len += 5);
228 *enc->cur++ = '\\'; 455 *enc->cur++ = '\\';
229 *enc->cur++ = 'u'; 456 *enc->cur++ = 'u';
230 *enc->cur++ = hexdigit [ uch >> 12 ]; 457 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
231 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 458 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
232 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 459 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
233 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 460 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
234 } 461 }
235 462
236 str += clen; 463 str += clen;
237 } 464 }
238 else if (enc->json.flags & F_LATIN1) 465 else if (enc->json.flags & F_LATIN1)
… …
250 while (--clen); 477 while (--clen);
251 } 478 }
252 else 479 else
253 { 480 {
254 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 481 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
255 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 482 enc->cur = encode_utf8 (enc->cur, uch);
256 ++str; 483 ++str;
257 } 484 }
258 } 485 }
259 } 486 }
260 } 487 }
261 488
262 --len; 489 --len;
263 } 490 }
264} 491}
265 492
266inline void 493INLINE void
267encode_indent (enc_t *enc) 494encode_indent (enc_t *enc)
268{ 495{
269 if (enc->json.flags & F_INDENT) 496 if (enc->json.flags & F_INDENT)
270 { 497 {
271 int spaces = enc->indent * INDENT_STEP; 498 int spaces = enc->indent * INDENT_STEP;
… …
274 memset (enc->cur, ' ', spaces); 501 memset (enc->cur, ' ', spaces);
275 enc->cur += spaces; 502 enc->cur += spaces;
276 } 503 }
277} 504}
278 505
279inline void 506INLINE void
280encode_space (enc_t *enc) 507encode_space (enc_t *enc)
281{ 508{
282 need (enc, 1); 509 need (enc, 1);
283 encode_ch (enc, ' '); 510 encode_ch (enc, ' ');
284} 511}
285 512
286inline void 513INLINE void
287encode_nl (enc_t *enc) 514encode_nl (enc_t *enc)
288{ 515{
289 if (enc->json.flags & F_INDENT) 516 if (enc->json.flags & F_INDENT)
290 { 517 {
291 need (enc, 1); 518 need (enc, 1);
292 encode_ch (enc, '\n'); 519 encode_ch (enc, '\n');
293 } 520 }
294} 521}
295 522
296inline void 523INLINE void
297encode_comma (enc_t *enc) 524encode_comma (enc_t *enc)
298{ 525{
299 encode_ch (enc, ','); 526 encode_ch (enc, ',');
300 527
301 if (enc->json.flags & F_INDENT) 528 if (enc->json.flags & F_INDENT)
… …
309static void 536static void
310encode_av (enc_t *enc, AV *av) 537encode_av (enc_t *enc, AV *av)
311{ 538{
312 int i, len = av_len (av); 539 int i, len = av_len (av);
313 540
314 if (enc->indent >= enc->maxdepth) 541 if (enc->indent >= enc->json.max_depth)
315 croak ("data structure too deep (hit recursion limit)"); 542 croak (ERR_NESTING_EXCEEDED);
316 543
317 encode_ch (enc, '['); encode_nl (enc); 544 encode_ch (enc, '[');
318 ++enc->indent;
319 545
546 if (len >= 0)
547 {
548 encode_nl (enc); ++enc->indent;
549
320 for (i = 0; i <= len; ++i) 550 for (i = 0; i <= len; ++i)
321 { 551 {
322 SV **svp = av_fetch (av, i, 0); 552 SV **svp = av_fetch (av, i, 0);
323 553
324 encode_indent (enc); 554 encode_indent (enc);
325 555
326 if (svp) 556 if (svp)
327 encode_sv (enc, *svp); 557 encode_sv (enc, *svp);
328 else 558 else
329 encode_str (enc, "null", 4, 0); 559 encode_str (enc, "null", 4, 0);
330 560
331 if (i < len) 561 if (i < len)
332 encode_comma (enc); 562 encode_comma (enc);
333 } 563 }
334 564
565 encode_nl (enc); --enc->indent; encode_indent (enc);
566 }
567
335 encode_nl (enc); 568 encode_ch (enc, ']');
336
337 --enc->indent;
338 encode_indent (enc); encode_ch (enc, ']');
339} 569}
340 570
341static void 571static void
342encode_hk (enc_t *enc, HE *he) 572encode_hk (enc_t *enc, HE *he)
343{ 573{
… …
346 if (HeKLEN (he) == HEf_SVKEY) 576 if (HeKLEN (he) == HEf_SVKEY)
347 { 577 {
348 SV *sv = HeSVKEY (he); 578 SV *sv = HeSVKEY (he);
349 STRLEN len; 579 STRLEN len;
350 char *str; 580 char *str;
351 581
352 SvGETMAGIC (sv); 582 SvGETMAGIC (sv);
353 str = SvPV (sv, len); 583 str = SvPV (sv, len);
354 584
355 encode_str (enc, str, len, SvUTF8 (sv)); 585 encode_str (enc, str, len, SvUTF8 (sv));
356 } 586 }
… …
391 621
392static void 622static void
393encode_hv (enc_t *enc, HV *hv) 623encode_hv (enc_t *enc, HV *hv)
394{ 624{
395 HE *he; 625 HE *he;
396 int count;
397 626
398 if (enc->indent >= enc->maxdepth) 627 if (enc->indent >= enc->json.max_depth)
399 croak ("data structure too deep (hit recursion limit)"); 628 croak (ERR_NESTING_EXCEEDED);
400 629
401 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 630 encode_ch (enc, '{');
402 631
403 // for canonical output we have to sort by keys first 632 // for canonical output we have to sort by keys first
404 // actually, this is mostly due to the stupid so-called 633 // actually, this is mostly due to the stupid so-called
405 // security workaround added somewhere in 5.8.x. 634 // security workaround added somewhere in 5.8.x
406 // that randomises hash orderings 635 // that randomises hash orderings
407 if (enc->json.flags & F_CANONICAL) 636 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
408 { 637 {
409 int count = hv_iterinit (hv); 638 int count = hv_iterinit (hv);
410 639
411 if (SvMAGICAL (hv)) 640 if (SvMAGICAL (hv))
412 { 641 {
… …
423 } 652 }
424 653
425 if (count) 654 if (count)
426 { 655 {
427 int i, fast = 1; 656 int i, fast = 1;
428#if defined(__BORLANDC__) || defined(_MSC_VER) 657 HE *hes_stack [STACK_HES];
429 HE **hes = _alloca (count * sizeof (HE)); 658 HE **hes = hes_stack;
430#else 659
431 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 660 // allocate larger arrays on the heap
432#endif 661 if (count > STACK_HES)
662 {
663 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
664 hes = (HE **)SvPVX (sv);
665 }
433 666
434 i = 0; 667 i = 0;
435 while ((he = hv_iternext (hv))) 668 while ((he = hv_iternext (hv)))
436 { 669 {
437 hes [i++] = he; 670 hes [i++] = he;
… …
459 692
460 FREETMPS; 693 FREETMPS;
461 LEAVE; 694 LEAVE;
462 } 695 }
463 696
697 encode_nl (enc); ++enc->indent;
698
464 while (count--) 699 while (count--)
465 { 700 {
466 encode_indent (enc); 701 encode_indent (enc);
467 he = hes [count]; 702 he = hes [count];
468 encode_hk (enc, he); 703 encode_hk (enc, he);
469 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); 704 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
470 705
471 if (count) 706 if (count)
472 encode_comma (enc); 707 encode_comma (enc);
473 } 708 }
709
710 encode_nl (enc); --enc->indent; encode_indent (enc);
474 } 711 }
475 } 712 }
476 else 713 else
477 { 714 {
478 if (hv_iterinit (hv) || SvMAGICAL (hv)) 715 if (hv_iterinit (hv) || SvMAGICAL (hv))
479 if ((he = hv_iternext (hv))) 716 if ((he = hv_iternext (hv)))
717 {
718 encode_nl (enc); ++enc->indent;
719
480 for (;;) 720 for (;;)
481 { 721 {
482 encode_indent (enc); 722 encode_indent (enc);
483 encode_hk (enc, he); 723 encode_hk (enc, he);
484 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); 724 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
485 725
486 if (!(he = hv_iternext (hv))) 726 if (!(he = hv_iternext (hv)))
487 break; 727 break;
488 728
489 encode_comma (enc); 729 encode_comma (enc);
490 } 730 }
491 }
492 731
732 encode_nl (enc); --enc->indent; encode_indent (enc);
733 }
734 }
735
493 encode_nl (enc); 736 encode_ch (enc, '}');
494
495 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
496} 737}
497 738
498// encode objects, arrays and special \0=false and \1=true values. 739// encode objects, arrays and special \0=false and \1=true values.
499static void 740static void
500encode_rv (enc_t *enc, SV *sv) 741encode_rv (enc_t *enc, SV *sv)
501{ 742{
502 svtype svt; 743 svtype svt;
744 GV *method;
503 745
504 SvGETMAGIC (sv); 746 SvGETMAGIC (sv);
505 svt = SvTYPE (sv); 747 svt = SvTYPE (sv);
506 748
507 if (expect_false (SvOBJECT (sv))) 749 if (expect_false (SvOBJECT (sv)))
508 { 750 {
509 HV *stash = !JSON_SLOW || json_boolean_stash 751 HV *stash = SvSTASH (sv);
510 ? json_boolean_stash
511 : gv_stashpv ("JSON::XS::Boolean", 1);
512 752
513 if (SvSTASH (sv) == stash) 753 if (stash == bool_stash)
514 { 754 {
515 if (SvIV (sv)) 755 if (SvIV (sv))
516 encode_str (enc, "true", 4, 0); 756 encode_str (enc, "true", 4, 0);
517 else 757 else
518 encode_str (enc, "false", 5, 0); 758 encode_str (enc, "false", 5, 0);
519 } 759 }
760 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
761 {
762 int count;
763 dSP;
764
765 ENTER; SAVETMPS;
766 SAVESTACK_POS ();
767 PUSHMARK (SP);
768 EXTEND (SP, 2);
769 // we re-bless the reference to get overload and other niceties right
770 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
771 PUSHs (sv_json);
772
773 PUTBACK;
774 count = call_sv ((SV *)GvCV (method), G_ARRAY);
775 SPAGAIN;
776
777 // catch this surprisingly common error
778 if (SvROK (TOPs) && SvRV (TOPs) == sv)
779 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
780
781 encode_ch (enc, '(');
782 encode_ch (enc, '"');
783 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
784 encode_ch (enc, '"');
785 encode_ch (enc, ')');
786 encode_ch (enc, '[');
787
788 while (count)
789 {
790 encode_sv (enc, SP[1 - count--]);
791
792 if (count)
793 encode_ch (enc, ',');
794 }
795
796 encode_ch (enc, ']');
797
798 FREETMPS; LEAVE;
799 }
800 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
801 {
802 dSP;
803
804 ENTER; SAVETMPS;
805 PUSHMARK (SP);
806 // we re-bless the reference to get overload and other niceties right
807 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
808
809 // calling with G_SCALAR ensures that we always get a 1 return value
810 PUTBACK;
811 call_sv ((SV *)GvCV (method), G_SCALAR);
812 SPAGAIN;
813
814 // catch this surprisingly common error
815 if (SvROK (TOPs) && SvRV (TOPs) == sv)
816 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
817
818 sv = POPs;
819 PUTBACK;
820
821 encode_sv (enc, sv);
822
823 FREETMPS; LEAVE;
824 }
825 else if (enc->json.flags & F_ALLOW_BLESSED)
826 encode_str (enc, "null", 4, 0);
520 else 827 else
521 {
522#if 0
523 if (0 && sv_derived_from (rv, "JSON::Literal"))
524 {
525 // not yet
526 }
527#endif
528 if (enc->json.flags & F_CONV_BLESSED)
529 {
530 // we re-bless the reference to get overload and other niceties right
531 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
532
533 if (to_json)
534 {
535 dSP;
536
537 ENTER; SAVETMPS; PUSHMARK (SP);
538 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
539
540 // calling with G_SCALAR ensures that we always get a 1 return value
541 PUTBACK;
542 call_sv ((SV *)GvCV (to_json), G_SCALAR);
543 SPAGAIN;
544
545 // catch this surprisingly common error
546 if (SvROK (TOPs) && SvRV (TOPs) == sv)
547 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
548
549 sv = POPs;
550 PUTBACK;
551
552 encode_sv (enc, sv);
553
554 FREETMPS; LEAVE;
555 }
556 else if (enc->json.flags & F_ALLOW_BLESSED)
557 encode_str (enc, "null", 4, 0);
558 else
559 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
560 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
561 }
562 else if (enc->json.flags & F_ALLOW_BLESSED)
563 encode_str (enc, "null", 4, 0);
564 else
565 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 828 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
566 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 829 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
567 }
568 } 830 }
569 else if (svt == SVt_PVHV) 831 else if (svt == SVt_PVHV)
570 encode_hv (enc, (HV *)sv); 832 encode_hv (enc, (HV *)sv);
571 else if (svt == SVt_PVAV) 833 else if (svt == SVt_PVAV)
572 encode_av (enc, (AV *)sv); 834 encode_av (enc, (AV *)sv);
573 else if (svt < SVt_PVAV) 835 else if (svt < SVt_PVAV)
574 { 836 {
575 STRLEN len = 0; 837 int bool_type = ref_bool_type (sv);
576 char *pv = svt ? SvPV (sv, len) : 0;
577 838
578 if (len == 1 && *pv == '1') 839 if (bool_type == 1)
579 encode_str (enc, "true", 4, 0); 840 encode_str (enc, "true", 4, 0);
580 else if (len == 1 && *pv == '0') 841 else if (bool_type == 0)
581 encode_str (enc, "false", 5, 0); 842 encode_str (enc, "false", 5, 0);
843 else if (enc->json.flags & F_ALLOW_UNKNOWN)
844 encode_str (enc, "null", 4, 0);
582 else 845 else
583 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 846 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
584 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 847 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
585 } 848 }
849 else if (enc->json.flags & F_ALLOW_UNKNOWN)
850 encode_str (enc, "null", 4, 0);
586 else 851 else
587 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 852 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
588 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 853 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
589} 854}
590 855
… …
608 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 873 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
609 enc->cur += strlen (enc->cur); 874 enc->cur += strlen (enc->cur);
610 } 875 }
611 else if (SvIOKp (sv)) 876 else if (SvIOKp (sv))
612 { 877 {
613 // we assume we can always read an IV as a UV 878 // we assume we can always read an IV as a UV and vice versa
614 if (SvUV (sv) & ~(UV)0x7fff) 879 // we assume two's complement
615 { 880 // we assume no aliasing issues in the union
616 // large integer, use the (rather slow) snprintf way. 881 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
617 need (enc, sizeof (UV) * 3); 882 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
618 enc->cur +=
619 SvIsUV(sv)
620 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
621 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
622 }
623 else
624 { 883 {
625 // optimise the "small number case" 884 // optimise the "small number case"
626 // code will likely be branchless and use only a single multiplication 885 // code will likely be branchless and use only a single multiplication
886 // works for numbers up to 59074
627 I32 i = SvIV (sv); 887 I32 i = SvIVX (sv);
628 U32 u; 888 U32 u;
629 char digit, nz = 0; 889 char digit, nz = 0;
630 890
631 need (enc, 6); 891 need (enc, 6);
632 892
… …
638 898
639 // now output digit by digit, each time masking out the integer part 899 // now output digit by digit, each time masking out the integer part
640 // and multiplying by 5 while moving the decimal point one to the right, 900 // and multiplying by 5 while moving the decimal point one to the right,
641 // resulting in a net multiplication by 10. 901 // resulting in a net multiplication by 10.
642 // we always write the digit to memory but conditionally increment 902 // we always write the digit to memory but conditionally increment
643 // the pointer, to ease the usage of conditional move instructions. 903 // the pointer, to enable the use of conditional move instructions.
644 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 904 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
645 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 905 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
646 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 906 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
647 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 907 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
648 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 908 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
649 } 909 }
910 else
911 {
912 // large integer, use the (rather slow) snprintf way.
913 need (enc, IVUV_MAXCHARS);
914 enc->cur +=
915 SvIsUV(sv)
916 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
917 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
918 }
650 } 919 }
651 else if (SvROK (sv)) 920 else if (SvROK (sv))
652 encode_rv (enc, SvRV (sv)); 921 encode_rv (enc, SvRV (sv));
653 else if (!SvOK (sv)) 922 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
654 encode_str (enc, "null", 4, 0); 923 encode_str (enc, "null", 4, 0);
655 else 924 else
656 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 925 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
657 SvPV_nolen (sv), SvFLAGS (sv)); 926 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
658} 927}
659 928
660static SV * 929static SV *
661encode_json (SV *scalar, JSON *json) 930encode_json (SV *scalar, JSON *json)
662{ 931{
663 enc_t enc; 932 enc_t enc;
664 933
665 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 934 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
666 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 935 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
667 936
668 enc.json = *json; 937 enc.json = *json;
669 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 938 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
670 enc.cur = SvPVX (enc.sv); 939 enc.cur = SvPVX (enc.sv);
671 enc.end = SvEND (enc.sv); 940 enc.end = SvEND (enc.sv);
672 enc.indent = 0; 941 enc.indent = 0;
673 enc.maxdepth = DEC_DEPTH (enc.json.flags); 942 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
943 : enc.json.flags & F_LATIN1 ? 0x000100UL
944 : 0x110000UL;
674 945
675 SvPOK_only (enc.sv); 946 SvPOK_only (enc.sv);
676 encode_sv (&enc, scalar); 947 encode_sv (&enc, scalar);
948 encode_nl (&enc);
677 949
678 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 950 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
679 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 951 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
680 952
681 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 953 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
699 JSON json; 971 JSON json;
700 U32 depth; // recursion depth 972 U32 depth; // recursion depth
701 U32 maxdepth; // recursion depth limit 973 U32 maxdepth; // recursion depth limit
702} dec_t; 974} dec_t;
703 975
704inline void 976INLINE void
977decode_comment (dec_t *dec)
978{
979 // only '#'-style comments allowed a.t.m.
980
981 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
982 ++dec->cur;
983}
984
985INLINE void
705decode_ws (dec_t *dec) 986decode_ws (dec_t *dec)
706{ 987{
707 for (;;) 988 for (;;)
708 { 989 {
709 char ch = *dec->cur; 990 char ch = *dec->cur;
710 991
711 if (ch > 0x20 992 if (ch > 0x20)
993 {
994 if (expect_false (ch == '#'))
995 {
996 if (dec->json.flags & F_RELAXED)
997 decode_comment (dec);
998 else
999 break;
1000 }
1001 else
1002 break;
1003 }
712 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 1004 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
713 break; 1005 break; // parse error, but let higher level handle it, gives better error messages
714
715 if (ch == '#' && dec->json.flags & F_RELAXED)
716 ++dec->cur;
717 1006
718 ++dec->cur; 1007 ++dec->cur;
719 } 1008 }
720} 1009}
721 1010
… …
725 if (*dec->cur != ch) \ 1014 if (*dec->cur != ch) \
726 ERR (# ch " expected"); \ 1015 ERR (# ch " expected"); \
727 ++dec->cur; \ 1016 ++dec->cur; \
728 SE 1017 SE
729 1018
730#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 1019#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
731#define DEC_DEC_DEPTH --dec->depth 1020#define DEC_DEC_DEPTH --dec->depth
732 1021
733static SV *decode_sv (dec_t *dec); 1022static SV *decode_sv (dec_t *dec);
734 1023
735static signed char decode_hexdigit[256]; 1024static signed char decode_hexdigit[256];
… …
827 1116
828 if (hi >= 0x80) 1117 if (hi >= 0x80)
829 { 1118 {
830 utf8 = 1; 1119 utf8 = 1;
831 1120
832 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1121 cur = encode_utf8 (cur, hi);
833 } 1122 }
834 else 1123 else
835 *cur++ = hi; 1124 *cur++ = hi;
836 } 1125 }
837 break; 1126 break;
… …
839 default: 1128 default:
840 --dec_cur; 1129 --dec_cur;
841 ERR ("illegal backslash escape sequence in string"); 1130 ERR ("illegal backslash escape sequence in string");
842 } 1131 }
843 } 1132 }
844 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1133 else if (expect_true (ch >= 0x20 && ch < 0x80))
845 *cur++ = ch; 1134 *cur++ = ch;
846 else if (ch >= 0x80) 1135 else if (ch >= 0x80)
847 { 1136 {
848 STRLEN clen; 1137 STRLEN clen;
849 UV uch;
850 1138
851 --dec_cur; 1139 --dec_cur;
852 1140
853 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1141 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
854 if (clen == (STRLEN)-1) 1142 if (clen == (STRLEN)-1)
855 ERR ("malformed UTF-8 character in JSON string"); 1143 ERR ("malformed UTF-8 character in JSON string");
856 1144
857 do 1145 do
858 *cur++ = *dec_cur++; 1146 *cur++ = *dec_cur++;
859 while (--clen); 1147 while (--clen);
860 1148
861 utf8 = 1; 1149 utf8 = 1;
862 } 1150 }
1151 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1152 *cur++ = ch;
863 else 1153 else
864 { 1154 {
865 --dec_cur; 1155 --dec_cur;
866 1156
867 if (!ch) 1157 if (!ch)
… …
875 { 1165 {
876 STRLEN len = cur - buf; 1166 STRLEN len = cur - buf;
877 1167
878 if (sv) 1168 if (sv)
879 { 1169 {
880 SvGROW (sv, SvCUR (sv) + len + 1); 1170 STRLEN cur = SvCUR (sv);
1171
1172 if (SvLEN (sv) <= cur + len)
1173 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1174
881 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1175 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
882 SvCUR_set (sv, SvCUR (sv) + len); 1176 SvCUR_set (sv, SvCUR (sv) + len);
883 } 1177 }
884 else 1178 else
885 sv = newSVpvn (buf, len); 1179 sv = newSVpvn (buf, len);
… …
972 1266
973 if (!is_nv) 1267 if (!is_nv)
974 { 1268 {
975 int len = dec->cur - start; 1269 int len = dec->cur - start;
976 1270
977 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1271 // special case the rather common 1..5-digit-int case
978 if (*start == '-') 1272 if (*start == '-')
979 switch (len) 1273 switch (len)
980 { 1274 {
981 case 2: return newSViv (-( start [1] - '0' * 1)); 1275 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
982 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1276 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
983 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1277 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
984 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1278 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1279 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
985 } 1280 }
986 else 1281 else
987 switch (len) 1282 switch (len)
988 { 1283 {
989 case 1: return newSViv ( start [0] - '0' * 1); 1284 case 1: return newSViv ( start [0] - '0' * 1);
990 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1285 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
991 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1286 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
992 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1287 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1288 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
993 } 1289 }
994 1290
995 { 1291 {
996 UV uv; 1292 UV uv;
997 int numtype = grok_number (start, len, &uv); 1293 int numtype = grok_number (start, len, &uv);
… …
1006 } 1302 }
1007 1303
1008 len -= *start == '-' ? 1 : 0; 1304 len -= *start == '-' ? 1 : 0;
1009 1305
1010 // does not fit into IV or UV, try NV 1306 // does not fit into IV or UV, try NV
1011 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1307 if (len <= NV_DIG)
1012 #if defined (LDBL_DIG)
1013 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1014 #endif
1015 )
1016 // fits into NV without loss of precision 1308 // fits into NV without loss of precision
1017 return newSVnv (Atof (start)); 1309 return newSVnv (json_atof (start));
1018 1310
1019 // everything else fails, convert it to a string 1311 // everything else fails, convert it to a string
1020 return newSVpvn (start, dec->cur - start); 1312 return newSVpvn (start, dec->cur - start);
1021 } 1313 }
1022 1314
1023 // loss of precision here 1315 // loss of precision here
1024 return newSVnv (Atof (start)); 1316 return newSVnv (json_atof (start));
1025 1317
1026fail: 1318fail:
1027 return 0; 1319 return 0;
1028} 1320}
1029 1321
… …
1053 if (*dec->cur == ']') 1345 if (*dec->cur == ']')
1054 { 1346 {
1055 ++dec->cur; 1347 ++dec->cur;
1056 break; 1348 break;
1057 } 1349 }
1058 1350
1059 if (*dec->cur != ',') 1351 if (*dec->cur != ',')
1060 ERR (", or ] expected while parsing array"); 1352 ERR (", or ] expected while parsing array");
1061 1353
1062 ++dec->cur; 1354 ++dec->cur;
1063 1355
… …
1105 char *p = dec->cur; 1397 char *p = dec->cur;
1106 char *e = p + 24; // only try up to 24 bytes 1398 char *e = p + 24; // only try up to 24 bytes
1107 1399
1108 for (;;) 1400 for (;;)
1109 { 1401 {
1110 // the >= 0x80 is true on most architectures 1402 // the >= 0x80 is false on most architectures
1111 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1403 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1112 { 1404 {
1113 // slow path, back up and use decode_str 1405 // slow path, back up and use decode_str
1114 SV *key = decode_str (dec); 1406 SV *key = decode_str (dec);
1115 if (!key) 1407 if (!key)
… …
1187 1479
1188 hv_iterinit (hv); 1480 hv_iterinit (hv);
1189 he = hv_iternext (hv); 1481 he = hv_iternext (hv);
1190 hv_iterinit (hv); 1482 hv_iterinit (hv);
1191 1483
1192 // the next line creates a mortal sv each time its called. 1484 // the next line creates a mortal sv each time it's called.
1193 // might want to optimise this for common cases. 1485 // might want to optimise this for common cases.
1194 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1486 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1195 1487
1196 if (cb) 1488 if (cb)
1197 { 1489 {
1198 dSP; 1490 dSP;
1199 int count; 1491 int count;
1200 1492
1201 ENTER; SAVETMPS; PUSHMARK (SP); 1493 ENTER; SAVETMPS;
1494 SAVESTACK_POS ();
1495 PUSHMARK (SP);
1202 XPUSHs (HeVAL (he)); 1496 XPUSHs (HeVAL (he));
1497 sv_2mortal (sv);
1203 1498
1204 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1499 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1205 1500
1206 if (count == 1) 1501 if (count == 1)
1207 { 1502 {
1208 sv = newSVsv (POPs); 1503 sv = newSVsv (POPs);
1209 FREETMPS; LEAVE; 1504 FREETMPS; LEAVE;
1210 return sv; 1505 return sv;
1211 } 1506 }
1212 1507
1508 SvREFCNT_inc (sv);
1213 FREETMPS; LEAVE; 1509 FREETMPS; LEAVE;
1214 } 1510 }
1215 } 1511 }
1216 1512
1217 if (dec->json.cb_object) 1513 if (dec->json.cb_object)
1218 { 1514 {
1219 dSP; 1515 dSP;
1220 int count; 1516 int count;
1221 1517
1222 ENTER; SAVETMPS; PUSHMARK (SP); 1518 ENTER; SAVETMPS;
1519 SAVESTACK_POS ();
1520 PUSHMARK (SP);
1223 XPUSHs (sv_2mortal (sv)); 1521 XPUSHs (sv_2mortal (sv));
1224 1522
1225 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1523 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1226 1524
1227 if (count == 1) 1525 if (count == 1)
… …
1243 DEC_DEC_DEPTH; 1541 DEC_DEC_DEPTH;
1244 return 0; 1542 return 0;
1245} 1543}
1246 1544
1247static SV * 1545static SV *
1546decode_tag (dec_t *dec)
1547{
1548 SV *tag = 0;
1549 SV *val = 0;
1550
1551 if (!(dec->json.flags & F_ALLOW_TAGS))
1552 ERR ("malformed JSON string, neither array, object, number, string or atom");
1553
1554 ++dec->cur;
1555
1556 decode_ws (dec);
1557
1558 tag = decode_sv (dec);
1559 if (!tag)
1560 goto fail;
1561
1562 if (!SvPOK (tag))
1563 ERR ("malformed JSON string, (tag) must be a string");
1564
1565 decode_ws (dec);
1566
1567 if (*dec->cur != ')')
1568 ERR (") expected after tag");
1569
1570 ++dec->cur;
1571
1572 decode_ws (dec);
1573
1574 val = decode_sv (dec);
1575 if (!val)
1576 goto fail;
1577
1578 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1579 ERR ("malformed JSON string, tag value must be an array");
1580
1581 {
1582 AV *av = (AV *)SvRV (val);
1583 int i, len = av_len (av) + 1;
1584 HV *stash = gv_stashsv (tag, 0);
1585 SV *sv;
1586
1587 if (!stash)
1588 ERR ("cannot decode perl-object (package does not exist)");
1589
1590 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1591
1592 if (!method)
1593 ERR ("cannot decode perl-object (package does not have a THAW method)");
1594
1595 dSP;
1596
1597 ENTER; SAVETMPS;
1598 PUSHMARK (SP);
1599 EXTEND (SP, len + 2);
1600 // we re-bless the reference to get overload and other niceties right
1601 PUSHs (tag);
1602 PUSHs (sv_json);
1603
1604 for (i = 0; i < len; ++i)
1605 PUSHs (*av_fetch (av, i, 1));
1606
1607 PUTBACK;
1608 call_sv ((SV *)GvCV (method), G_SCALAR);
1609 SPAGAIN;
1610
1611 SvREFCNT_dec (tag);
1612 SvREFCNT_dec (val);
1613 sv = SvREFCNT_inc (POPs);
1614
1615 PUTBACK;
1616
1617 FREETMPS; LEAVE;
1618
1619 return sv;
1620 }
1621
1622fail:
1623 SvREFCNT_dec (tag);
1624 SvREFCNT_dec (val);
1625 return 0;
1626}
1627
1628static SV *
1248decode_sv (dec_t *dec) 1629decode_sv (dec_t *dec)
1249{ 1630{
1250 // the beauty of JSON: you need exactly one character lookahead 1631 // the beauty of JSON: you need exactly one character lookahead
1251 // to parse anything. 1632 // to parse everything.
1252 switch (*dec->cur) 1633 switch (*dec->cur)
1253 { 1634 {
1254 case '"': ++dec->cur; return decode_str (dec); 1635 case '"': ++dec->cur; return decode_str (dec);
1255 case '[': ++dec->cur; return decode_av (dec); 1636 case '[': ++dec->cur; return decode_av (dec);
1256 case '{': ++dec->cur; return decode_hv (dec); 1637 case '{': ++dec->cur; return decode_hv (dec);
1638 case '(': return decode_tag (dec);
1257 1639
1258 case '-': 1640 case '-':
1259 case '0': case '1': case '2': case '3': case '4': 1641 case '0': case '1': case '2': case '3': case '4':
1260 case '5': case '6': case '7': case '8': case '9': 1642 case '5': case '6': case '7': case '8': case '9':
1261 return decode_num (dec); 1643 return decode_num (dec);
… …
1263 case 't': 1645 case 't':
1264 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1646 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1265 { 1647 {
1266 dec->cur += 4; 1648 dec->cur += 4;
1267#if JSON_SLOW 1649#if JSON_SLOW
1268 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1650 bool_true = get_bool ("Types::Serialiser::true");
1269#endif 1651#endif
1270 return SvREFCNT_inc (json_true); 1652 return newSVsv (bool_true);
1271 } 1653 }
1272 else 1654 else
1273 ERR ("'true' expected"); 1655 ERR ("'true' expected");
1274 1656
1275 break; 1657 break;
… …
1277 case 'f': 1659 case 'f':
1278 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1660 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1279 { 1661 {
1280 dec->cur += 5; 1662 dec->cur += 5;
1281#if JSON_SLOW 1663#if JSON_SLOW
1282 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1664 bool_false = get_bool ("Types::Serialiser::false");
1283#endif 1665#endif
1284 return SvREFCNT_inc (json_false); 1666 return newSVsv (bool_false);
1285 } 1667 }
1286 else 1668 else
1287 ERR ("'false' expected"); 1669 ERR ("'false' expected");
1288 1670
1289 break; 1671 break;
… …
1298 ERR ("'null' expected"); 1680 ERR ("'null' expected");
1299 1681
1300 break; 1682 break;
1301 1683
1302 default: 1684 default:
1303 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1685 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1304 break; 1686 break;
1305 } 1687 }
1306 1688
1307fail: 1689fail:
1308 return 0; 1690 return 0;
1309} 1691}
1310 1692
1311static SV * 1693static SV *
1312decode_json (SV *string, JSON *json, UV *offset_return) 1694decode_json (SV *string, JSON *json, STRLEN *offset_return)
1313{ 1695{
1314 dec_t dec; 1696 dec_t dec;
1315 UV offset;
1316 SV *sv; 1697 SV *sv;
1317 1698
1699 /* work around bugs in 5.10 where manipulating magic values
1700 * makes perl ignore the magic in subsequent accesses.
1701 * also make a copy of non-PV values, to get them into a clean
1702 * state (SvPV should do that, but it's buggy, see below).
1703 *
1704 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1705 * as reported by Reini Urban.
1706 */
1318 SvGETMAGIC (string); 1707 /*SvGETMAGIC (string);*/
1708 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1709 string = sv_2mortal (newSVsv (string));
1710
1319 SvUPGRADE (string, SVt_PV); 1711 SvUPGRADE (string, SVt_PV);
1320 1712
1321 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1713 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1714 * assertion with -DDEBUGGING, although SvCUR is documented to
1715 * return the xpv_cur field which certainly exists after upgrading.
1716 * according to nicholas clark, calling SvPOK fixes this.
1717 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1718 * and hope for the best.
1719 * Damnit, SvPV_nolen still trips over yet another assertion. This
1720 * assertion business is seriously broken, try yet another workaround
1721 * for the broken -DDEBUGGING.
1722 */
1723 {
1724#ifdef DEBUGGING
1725 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1726#else
1727 STRLEN offset = SvCUR (string);
1728#endif
1729
1730 if (offset > json->max_size && json->max_size)
1322 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1731 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1323 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1732 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1733 }
1324 1734
1325 if (json->flags & F_UTF8) 1735 if (DECODE_WANTS_OCTETS (json))
1326 sv_utf8_downgrade (string, 0); 1736 sv_utf8_downgrade (string, 0);
1327 else 1737 else
1328 sv_utf8_upgrade (string); 1738 sv_utf8_upgrade (string);
1329 1739
1330 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1740 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1331 1741
1332 dec.json = *json; 1742 dec.json = *json;
1333 dec.cur = SvPVX (string); 1743 dec.cur = SvPVX (string);
1334 dec.end = SvEND (string); 1744 dec.end = SvEND (string);
1335 dec.err = 0; 1745 dec.err = 0;
1336 dec.depth = 0; 1746 dec.depth = 0;
1337 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1338 1747
1339 if (dec.json.cb_object || dec.json.cb_sk_object) 1748 if (dec.json.cb_object || dec.json.cb_sk_object)
1340 dec.json.flags |= F_HOOK; 1749 dec.json.flags |= F_HOOK;
1341 1750
1342 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1751 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1343 1752
1344 decode_ws (&dec); 1753 decode_ws (&dec);
1345 sv = decode_sv (&dec); 1754 sv = decode_sv (&dec);
1346 1755
1347 if (!(offset_return || !sv)) 1756 if (offset_return)
1757 *offset_return = dec.cur - SvPVX (string);
1758 else if (sv)
1348 { 1759 {
1349 // check for trailing garbage 1760 // check for trailing garbage
1350 decode_ws (&dec); 1761 decode_ws (&dec);
1351 1762
1352 if (*dec.cur) 1763 if (*dec.cur)
1353 { 1764 {
1354 dec.err = "garbage after JSON object"; 1765 dec.err = "garbage after JSON object";
1355 SvREFCNT_dec (sv); 1766 SvREFCNT_dec (sv);
1356 sv = 0; 1767 sv = 0;
1357 } 1768 }
1358 }
1359
1360 if (offset_return || !sv)
1361 {
1362 offset = dec.json.flags & F_UTF8
1363 ? dec.cur - SvPVX (string)
1364 : utf8_distance (dec.cur, SvPVX (string));
1365
1366 if (offset_return)
1367 *offset_return = offset;
1368 } 1769 }
1369 1770
1370 if (!sv) 1771 if (!sv)
1371 { 1772 {
1372 SV *uni = sv_newmortal (); 1773 SV *uni = sv_newmortal ();
… …
1378 SAVEVPTR (PL_curcop); 1779 SAVEVPTR (PL_curcop);
1379 PL_curcop = &cop; 1780 PL_curcop = &cop;
1380 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1781 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1381 LEAVE; 1782 LEAVE;
1382 1783
1383 croak ("%s, at character offset %d [\"%s\"]", 1784 croak ("%s, at character offset %d (before \"%s\")",
1384 dec.err, 1785 dec.err,
1385 (int)offset, 1786 (int)ptr_to_index (string, dec.cur),
1386 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1787 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1387 } 1788 }
1388 1789
1389 sv = sv_2mortal (sv); 1790 sv = sv_2mortal (sv);
1390 1791
1391 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1792 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1392 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1793 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1393 1794
1394 return sv; 1795 return sv;
1796}
1797
1798/////////////////////////////////////////////////////////////////////////////
1799// incremental parser
1800
1801static void
1802incr_parse (JSON *self)
1803{
1804 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1805
1806 // the state machine here is a bit convoluted and could be simplified a lot
1807 // but this would make it slower, so...
1808
1809 for (;;)
1810 {
1811 //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
1812 switch (self->incr_mode)
1813 {
1814 // only used for initial whitespace skipping
1815 case INCR_M_WS:
1816 for (;;)
1817 {
1818 if (*p > 0x20)
1819 {
1820 if (*p == '#')
1821 {
1822 self->incr_mode = INCR_M_C0;
1823 goto incr_m_c;
1824 }
1825 else
1826 {
1827 self->incr_mode = INCR_M_JSON;
1828 goto incr_m_json;
1829 }
1830 }
1831 else if (!*p)
1832 goto interrupt;
1833
1834 ++p;
1835 }
1836
1837 // skip a single char inside a string (for \\-processing)
1838 case INCR_M_BS:
1839 if (!*p)
1840 goto interrupt;
1841
1842 ++p;
1843 self->incr_mode = INCR_M_STR;
1844 goto incr_m_str;
1845
1846 // inside #-style comments
1847 case INCR_M_C0:
1848 case INCR_M_C1:
1849 incr_m_c:
1850 for (;;)
1851 {
1852 if (*p == '\n')
1853 {
1854 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1855 break;
1856 }
1857 else if (!*p)
1858 goto interrupt;
1859
1860 ++p;
1861 }
1862
1863 break;
1864
1865 // inside a string
1866 case INCR_M_STR:
1867 incr_m_str:
1868 for (;;)
1869 {
1870 if (*p == '"')
1871 {
1872 ++p;
1873 self->incr_mode = INCR_M_JSON;
1874
1875 if (!self->incr_nest)
1876 goto interrupt;
1877
1878 goto incr_m_json;
1879 }
1880 else if (*p == '\\')
1881 {
1882 ++p; // "virtually" consumes character after \
1883
1884 if (!*p) // if at end of string we have to switch modes
1885 {
1886 self->incr_mode = INCR_M_BS;
1887 goto interrupt;
1888 }
1889 }
1890 else if (!*p)
1891 goto interrupt;
1892
1893 ++p;
1894 }
1895
1896 // after initial ws, outside string
1897 case INCR_M_JSON:
1898 incr_m_json:
1899 for (;;)
1900 {
1901 switch (*p++)
1902 {
1903 case 0:
1904 --p;
1905 goto interrupt;
1906
1907 case 0x09:
1908 case 0x0a:
1909 case 0x0d:
1910 case 0x20:
1911 if (!self->incr_nest)
1912 {
1913 --p; // do not eat the whitespace, let the next round do it
1914 goto interrupt;
1915 }
1916 break;
1917
1918 case '"':
1919 self->incr_mode = INCR_M_STR;
1920 goto incr_m_str;
1921
1922 case '[':
1923 case '{':
1924 case '(':
1925 if (++self->incr_nest > self->max_depth)
1926 croak (ERR_NESTING_EXCEEDED);
1927 break;
1928
1929 case ']':
1930 case '}':
1931 if (--self->incr_nest <= 0)
1932 goto interrupt;
1933 break;
1934
1935 case ')':
1936 --self->incr_nest;
1937 break;
1938
1939 case '#':
1940 self->incr_mode = INCR_M_C1;
1941 goto incr_m_c;
1942 }
1943 }
1944 }
1945
1946 modechange:
1947 ;
1948 }
1949
1950interrupt:
1951 self->incr_pos = p - SvPVX (self->incr_text);
1952 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1953 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1395} 1954}
1396 1955
1397///////////////////////////////////////////////////////////////////////////// 1956/////////////////////////////////////////////////////////////////////////////
1398// XS interface functions 1957// XS interface functions
1399 1958
… …
1408 i >= '0' && i <= '9' ? i - '0' 1967 i >= '0' && i <= '9' ? i - '0'
1409 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1968 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1410 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1969 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1411 : -1; 1970 : -1;
1412 1971
1413 json_stash = gv_stashpv ("JSON::XS" , 1); 1972 json_stash = gv_stashpv ("JSON::XS" , 1);
1414 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1973 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1974 bool_true = get_bool ("Types::Serialiser::true");
1975 bool_false = get_bool ("Types::Serialiser::false");
1415 1976
1416 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1977 sv_json = newSVpv ("JSON", 0);
1417 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1978 SvREADONLY_on (sv_json);
1979
1980 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1418} 1981}
1419 1982
1420PROTOTYPES: DISABLE 1983PROTOTYPES: DISABLE
1421 1984
1422void CLONE (...) 1985void CLONE (...)
1423 CODE: 1986 CODE:
1424 json_stash = 0; 1987 json_stash = 0;
1425 json_boolean_stash = 0; 1988 bool_stash = 0;
1426 1989
1427void new (char *klass) 1990void new (char *klass)
1428 PPCODE: 1991 PPCODE:
1429{ 1992{
1430 SV *pv = NEWSV (0, sizeof (JSON)); 1993 SV *pv = NEWSV (0, sizeof (JSON));
1431 SvPOK_only (pv); 1994 SvPOK_only (pv);
1432 Zero (SvPVX (pv), 1, JSON); 1995 json_init ((JSON *)SvPVX (pv));
1433 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1996 XPUSHs (sv_2mortal (sv_bless (
1434 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH))); 1997 newRV_noinc (pv),
1998 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1999 )));
1435} 2000}
1436 2001
1437void ascii (JSON *self, int enable = 1) 2002void ascii (JSON *self, int enable = 1)
1438 ALIAS: 2003 ALIAS:
1439 ascii = F_ASCII 2004 ascii = F_ASCII
… …
1447 allow_nonref = F_ALLOW_NONREF 2012 allow_nonref = F_ALLOW_NONREF
1448 shrink = F_SHRINK 2013 shrink = F_SHRINK
1449 allow_blessed = F_ALLOW_BLESSED 2014 allow_blessed = F_ALLOW_BLESSED
1450 convert_blessed = F_CONV_BLESSED 2015 convert_blessed = F_CONV_BLESSED
1451 relaxed = F_RELAXED 2016 relaxed = F_RELAXED
2017 allow_unknown = F_ALLOW_UNKNOWN
2018 allow_tags = F_ALLOW_TAGS
1452 PPCODE: 2019 PPCODE:
1453{ 2020{
1454 if (enable) 2021 if (enable)
1455 self->flags |= ix; 2022 self->flags |= ix;
1456 else 2023 else
1457 self->flags &= ~ix; 2024 self->flags &= ~ix;
1458 2025
1459 XPUSHs (ST (0)); 2026 XPUSHs (ST (0));
1460} 2027}
1461 2028
1462void max_depth (JSON *self, UV max_depth = 0x80000000UL) 2029void get_ascii (JSON *self)
2030 ALIAS:
2031 get_ascii = F_ASCII
2032 get_latin1 = F_LATIN1
2033 get_utf8 = F_UTF8
2034 get_indent = F_INDENT
2035 get_canonical = F_CANONICAL
2036 get_space_before = F_SPACE_BEFORE
2037 get_space_after = F_SPACE_AFTER
2038 get_allow_nonref = F_ALLOW_NONREF
2039 get_shrink = F_SHRINK
2040 get_allow_blessed = F_ALLOW_BLESSED
2041 get_convert_blessed = F_CONV_BLESSED
2042 get_relaxed = F_RELAXED
2043 get_allow_unknown = F_ALLOW_UNKNOWN
2044 get_allow_tags = F_ALLOW_TAGS
1463 PPCODE: 2045 PPCODE:
1464{ 2046 XPUSHs (boolSV (self->flags & ix));
1465 UV log2 = 0;
1466 2047
1467 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 2048void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1468 2049 PPCODE:
1469 while ((1UL << log2) < max_depth) 2050 self->max_depth = max_depth;
1470 ++log2;
1471
1472 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1473
1474 XPUSHs (ST (0)); 2051 XPUSHs (ST (0));
1475}
1476 2052
2053U32 get_max_depth (JSON *self)
2054 CODE:
2055 RETVAL = self->max_depth;
2056 OUTPUT:
2057 RETVAL
2058
1477void max_size (JSON *self, UV max_size = 0) 2059void max_size (JSON *self, U32 max_size = 0)
1478 PPCODE: 2060 PPCODE:
1479{ 2061 self->max_size = max_size;
1480 UV log2 = 0;
1481
1482 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1483 if (max_size == 1) max_size = 2;
1484
1485 while ((1UL << log2) < max_size)
1486 ++log2;
1487
1488 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1489
1490 XPUSHs (ST (0)); 2062 XPUSHs (ST (0));
1491} 2063
2064int get_max_size (JSON *self)
2065 CODE:
2066 RETVAL = self->max_size;
2067 OUTPUT:
2068 RETVAL
1492 2069
1493void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 2070void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1494 PPCODE: 2071 PPCODE:
1495{ 2072{
1496 SvREFCNT_dec (self->cb_object); 2073 SvREFCNT_dec (self->cb_object);
… …
1500} 2077}
1501 2078
1502void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2079void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1503 PPCODE: 2080 PPCODE:
1504{ 2081{
1505 if (!self->cb_sk_object) 2082 if (!self->cb_sk_object)
1506 self->cb_sk_object = newHV (); 2083 self->cb_sk_object = newHV ();
1507 2084
1508 if (SvOK (cb)) 2085 if (SvOK (cb))
1509 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2086 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1510 else 2087 else
… …
1521 XPUSHs (ST (0)); 2098 XPUSHs (ST (0));
1522} 2099}
1523 2100
1524void encode (JSON *self, SV *scalar) 2101void encode (JSON *self, SV *scalar)
1525 PPCODE: 2102 PPCODE:
1526 XPUSHs (encode_json (scalar, self)); 2103 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2104 XPUSHs (scalar);
1527 2105
1528void decode (JSON *self, SV *jsonstr) 2106void decode (JSON *self, SV *jsonstr)
1529 PPCODE: 2107 PPCODE:
1530 XPUSHs (decode_json (jsonstr, self, 0)); 2108 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2109 XPUSHs (jsonstr);
1531 2110
1532void decode_prefix (JSON *self, SV *jsonstr) 2111void decode_prefix (JSON *self, SV *jsonstr)
1533 PPCODE: 2112 PPCODE:
1534{ 2113{
1535 UV offset; 2114 SV *sv;
2115 STRLEN offset;
2116 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1536 EXTEND (SP, 2); 2117 EXTEND (SP, 2);
1537 PUSHs (decode_json (jsonstr, self, &offset)); 2118 PUSHs (sv);
1538 PUSHs (sv_2mortal (newSVuv (offset))); 2119 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
2120}
2121
2122void incr_parse (JSON *self, SV *jsonstr = 0)
2123 PPCODE:
2124{
2125 if (!self->incr_text)
2126 self->incr_text = newSVpvn ("", 0);
2127
2128 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2129 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2130 if (DECODE_WANTS_OCTETS (self))
2131 {
2132 if (self->incr_pos)
2133 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2134 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2135
2136 sv_utf8_downgrade (self->incr_text, 0);
2137 }
2138 else
2139 {
2140 sv_utf8_upgrade (self->incr_text);
2141
2142 if (self->incr_pos)
2143 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2144 - (U8 *)SvPVX (self->incr_text);
2145 }
2146
2147 // append data, if any
2148 if (jsonstr)
2149 {
2150 /* make sure both strings have same encoding */
2151 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2152 if (SvUTF8 (jsonstr))
2153 sv_utf8_downgrade (jsonstr, 0);
2154 else
2155 sv_utf8_upgrade (jsonstr);
2156
2157 /* and then just blindly append */
2158 {
2159 STRLEN len;
2160 const char *str = SvPV (jsonstr, len);
2161 STRLEN cur = SvCUR (self->incr_text);
2162
2163 if (SvLEN (self->incr_text) <= cur + len)
2164 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
2165
2166 Move (str, SvEND (self->incr_text), len, char);
2167 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2168 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2169 }
2170 }
2171
2172 if (GIMME_V != G_VOID)
2173 do
2174 {
2175 SV *sv;
2176 STRLEN offset;
2177
2178 if (!INCR_DONE (self))
2179 {
2180 incr_parse (self);
2181
2182 if (self->incr_pos > self->max_size && self->max_size)
2183 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2184 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2185
2186 if (!INCR_DONE (self))
2187 {
2188 // as an optimisation, do not accumulate white space in the incr buffer
2189 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2190 {
2191 self->incr_pos = 0;
2192 SvCUR_set (self->incr_text, 0);
2193 }
2194
2195 break;
2196 }
2197 }
2198
2199 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2200 XPUSHs (sv);
2201
2202 self->incr_pos -= offset;
2203 self->incr_nest = 0;
2204 self->incr_mode = 0;
2205
2206 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
2207 }
2208 while (GIMME_V == G_ARRAY);
2209}
2210
2211SV *incr_text (JSON *self)
2212 ATTRS: lvalue
2213 CODE:
2214{
2215 if (self->incr_pos)
2216 croak ("incr_text can not be called when the incremental parser already started parsing");
2217
2218 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2219}
2220 OUTPUT:
2221 RETVAL
2222
2223void incr_skip (JSON *self)
2224 CODE:
2225{
2226 if (self->incr_pos)
2227 {
2228 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2229 self->incr_pos = 0;
2230 self->incr_nest = 0;
2231 self->incr_mode = 0;
2232 }
2233}
2234
2235void incr_reset (JSON *self)
2236 CODE:
2237{
2238 SvREFCNT_dec (self->incr_text);
2239 self->incr_text = 0;
2240 self->incr_pos = 0;
2241 self->incr_nest = 0;
2242 self->incr_mode = 0;
1539} 2243}
1540 2244
1541void DESTROY (JSON *self) 2245void DESTROY (JSON *self)
1542 CODE: 2246 CODE:
1543 SvREFCNT_dec (self->cb_sk_object); 2247 SvREFCNT_dec (self->cb_sk_object);
1544 SvREFCNT_dec (self->cb_object); 2248 SvREFCNT_dec (self->cb_object);
2249 SvREFCNT_dec (self->incr_text);
1545 2250
1546PROTOTYPES: ENABLE 2251PROTOTYPES: ENABLE
1547 2252
1548void to_json (SV *scalar) 2253void encode_json (SV *scalar)
1549 PPCODE: 2254 PPCODE:
1550{ 2255{
1551 JSON json = { F_DEFAULT | F_UTF8 }; 2256 JSON json;
1552 XPUSHs (encode_json (scalar, &json)); 2257 json_init (&json);
2258 json.flags |= F_UTF8;
2259 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2260 XPUSHs (scalar);
1553} 2261}
1554 2262
1555void from_json (SV *jsonstr) 2263void decode_json (SV *jsonstr)
1556 PPCODE: 2264 PPCODE:
1557{ 2265{
1558 JSON json = { F_DEFAULT | F_UTF8 }; 2266 JSON json;
1559 XPUSHs (decode_json (jsonstr, &json, 0)); 2267 json_init (&json);
2268 json.flags |= F_UTF8;
2269 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2270 XPUSHs (jsonstr);
1560} 2271}
1561 2272

Diff Legend

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