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 1.66 by root, Sun Nov 25 19:02:42 2007 UTC vs.
Revision 1.124 by root, Mon Nov 24 18:42:51 2014 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))
68#else 84#else
69# define JSON_SLOW 0 85# define JSON_SLOW 0
70# define JSON_STASH json_stash 86# define JSON_STASH json_stash
71#endif 87#endif
72 88
89// the amount of HEs to allocate on the stack, when sorting keys
90#define STACK_HES 64
91
73static HV *json_stash, *json_boolean_stash; // JSON::XS:: 92static HV *json_stash, *types_boolean_stash; // JSON::XS::
74static SV *json_true, *json_false; 93static SV *types_true, *types_false, *sv_json;
94
95enum {
96 INCR_M_WS = 0, // initial whitespace skipping, must be 0
97 INCR_M_STR, // inside string
98 INCR_M_BS, // inside backslash
99 INCR_M_C0, // inside comment in initial whitespace sequence
100 INCR_M_C1, // inside comment in other places
101 INCR_M_JSON // outside anything, count nesting
102};
103
104#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
75 105
76typedef struct { 106typedef struct {
77 U32 flags; 107 U32 flags;
108 U32 max_depth;
109 STRLEN max_size;
110
78 SV *cb_object; 111 SV *cb_object;
79 HV *cb_sk_object; 112 HV *cb_sk_object;
113
114 // for the incremental parser
115 SV *incr_text; // the source text so far
116 STRLEN incr_pos; // the current offset into the text
117 int incr_nest; // {[]}-nesting level
118 unsigned char incr_mode;
80} JSON; 119} JSON;
120
121INLINE void
122json_init (JSON *json)
123{
124 Zero (json, 1, JSON);
125 json->max_depth = 512;
126}
81 127
82///////////////////////////////////////////////////////////////////////////// 128/////////////////////////////////////////////////////////////////////////////
83// utility functions 129// utility functions
84 130
85inline void 131INLINE SV *
132get_bool (const char *name)
133{
134 SV *sv = get_sv (name, 1);
135
136 SvREADONLY_on (sv);
137 SvREADONLY_on (SvRV (sv));
138
139 return sv;
140}
141
142INLINE void
86shrink (SV *sv) 143shrink (SV *sv)
87{ 144{
88 sv_utf8_downgrade (sv, 1); 145 sv_utf8_downgrade (sv, 1);
146
89 if (SvLEN (sv) > SvCUR (sv) + 1) 147 if (SvLEN (sv) > SvCUR (sv) + 1)
90 { 148 {
91#ifdef SvPV_shrink_to_cur 149#ifdef SvPV_shrink_to_cur
92 SvPV_shrink_to_cur (sv); 150 SvPV_shrink_to_cur (sv);
93#elif defined (SvPV_renew) 151#elif defined (SvPV_renew)
… …
99// decode an utf-8 character and return it, or (UV)-1 in 157// decode an utf-8 character and return it, or (UV)-1 in
100// case of an error. 158// case of an error.
101// we special-case "safe" characters from U+80 .. U+7FF, 159// we special-case "safe" characters from U+80 .. U+7FF,
102// but use the very good perl function to parse anything else. 160// but use the very good perl function to parse anything else.
103// note that we never call this function for a ascii codepoints 161// note that we never call this function for a ascii codepoints
104inline UV 162INLINE UV
105decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 163decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
106{ 164{
107 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 165 if (expect_true (len >= 2
108 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 166 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
109 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 167 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
110 { 168 {
111 *clen = 2; 169 *clen = 2;
112 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 170 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
113 } 171 }
114 else 172 else
115 { 173 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
116 *clen = (STRLEN)-1; 174}
117 return (UV)-1; 175
176// likewise for encoding, also never called for ascii codepoints
177// this function takes advantage of this fact, although current gccs
178// seem to optimise the check for >= 0x80 away anyways
179INLINE unsigned char *
180encode_utf8 (unsigned char *s, UV ch)
181{
182 if (expect_false (ch < 0x000080))
183 *s++ = ch;
184 else if (expect_true (ch < 0x000800))
185 *s++ = 0xc0 | ( ch >> 6),
186 *s++ = 0x80 | ( ch & 0x3f);
187 else if ( ch < 0x010000)
188 *s++ = 0xe0 | ( ch >> 12),
189 *s++ = 0x80 | ((ch >> 6) & 0x3f),
190 *s++ = 0x80 | ( ch & 0x3f);
191 else if ( ch < 0x110000)
192 *s++ = 0xf0 | ( ch >> 18),
193 *s++ = 0x80 | ((ch >> 12) & 0x3f),
194 *s++ = 0x80 | ((ch >> 6) & 0x3f),
195 *s++ = 0x80 | ( ch & 0x3f);
196
197 return s;
198}
199
200// convert offset pointer to character index, sv must be string
201static STRLEN
202ptr_to_index (SV *sv, char *offset)
203{
204 return SvUTF8 (sv)
205 ? utf8_distance (offset, SvPVX (sv))
206 : offset - SvPVX (sv);
207}
208
209/////////////////////////////////////////////////////////////////////////////
210// fp hell
211
212// scan a group of digits, and a trailing exponent
213static void
214json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
215{
216 UV uaccum = 0;
217 int eaccum = 0;
218
219 // if we recurse too deep, skip all remaining digits
220 // to avoid a stack overflow attack
221 if (expect_false (--maxdepth <= 0))
222 while (((U8)*s - '0') < 10)
223 ++s;
224
225 for (;;)
118 } 226 {
119} 227 U8 dig = (U8)*s - '0';
120 228
229 if (expect_false (dig >= 10))
230 {
231 if (dig == (U8)((U8)'.' - (U8)'0'))
232 {
233 ++s;
234 json_atof_scan1 (s, accum, expo, 1, maxdepth);
235 }
236 else if ((dig | ' ') == 'e' - '0')
237 {
238 int exp2 = 0;
239 int neg = 0;
240
241 ++s;
242
243 if (*s == '-')
244 {
245 ++s;
246 neg = 1;
247 }
248 else if (*s == '+')
249 ++s;
250
251 while ((dig = (U8)*s - '0') < 10)
252 exp2 = exp2 * 10 + *s++ - '0';
253
254 *expo += neg ? -exp2 : exp2;
255 }
256
257 break;
258 }
259
260 ++s;
261
262 uaccum = uaccum * 10 + dig;
263 ++eaccum;
264
265 // if we have too many digits, then recurse for more
266 // we actually do this for rather few digits
267 if (uaccum >= (UV_MAX - 9) / 10)
268 {
269 if (postdp) *expo -= eaccum;
270 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
271 if (postdp) *expo += eaccum;
272
273 break;
274 }
275 }
276
277 // this relies greatly on the quality of the pow ()
278 // implementation of the platform, but a good
279 // implementation is hard to beat.
280 // (IEEE 754 conformant ones are required to be exact)
281 if (postdp) *expo -= eaccum;
282 *accum += uaccum * Perl_pow (10., *expo);
283 *expo += eaccum;
284}
285
286static NV
287json_atof (const char *s)
288{
289 NV accum = 0.;
290 int expo = 0;
291 int neg = 0;
292
293 if (*s == '-')
294 {
295 ++s;
296 neg = 1;
297 }
298
299 // a recursion depth of ten gives us >>500 bits
300 json_atof_scan1 (s, &accum, &expo, 0, 10);
301
302 return neg ? -accum : accum;
303}
121///////////////////////////////////////////////////////////////////////////// 304/////////////////////////////////////////////////////////////////////////////
122// encoder 305// encoder
123 306
124// structure used for encoding JSON 307// structure used for encoding JSON
125typedef struct 308typedef struct
… …
127 char *cur; // SvPVX (sv) + current output position 310 char *cur; // SvPVX (sv) + current output position
128 char *end; // SvEND (sv) 311 char *end; // SvEND (sv)
129 SV *sv; // result scalar 312 SV *sv; // result scalar
130 JSON json; 313 JSON json;
131 U32 indent; // indentation level 314 U32 indent; // indentation level
132 U32 maxdepth; // max. indentation/recursion level 315 UV limit; // escape character values >= this value when encoding
133} enc_t; 316} enc_t;
134 317
135inline void 318INLINE void
136need (enc_t *enc, STRLEN len) 319need (enc_t *enc, STRLEN len)
137{ 320{
138 if (expect_false (enc->cur + len >= enc->end)) 321 if (expect_false (enc->cur + len >= enc->end))
139 { 322 {
140 STRLEN cur = enc->cur - SvPVX (enc->sv); 323 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
141 SvGROW (enc->sv, cur + len + 1); 324 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
142 enc->cur = SvPVX (enc->sv) + cur; 325 enc->cur = SvPVX (enc->sv) + cur;
143 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 326 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
144 } 327 }
145} 328}
146 329
147inline void 330INLINE void
148encode_ch (enc_t *enc, char ch) 331encode_ch (enc_t *enc, char ch)
149{ 332{
150 need (enc, 1); 333 need (enc, 1);
151 *enc->cur++ = ch; 334 *enc->cur++ = ch;
152} 335}
… …
206 { 389 {
207 uch = ch; 390 uch = ch;
208 clen = 1; 391 clen = 1;
209 } 392 }
210 393
211 if (uch > 0x10FFFFUL) 394 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 { 395 {
216 if (uch > 0xFFFFUL) 396 if (uch >= 0x10000UL)
217 { 397 {
398 if (uch >= 0x110000UL)
399 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
400
218 need (enc, len += 11); 401 need (enc, len += 11);
219 sprintf (enc->cur, "\\u%04x\\u%04x", 402 sprintf (enc->cur, "\\u%04x\\u%04x",
220 (int)((uch - 0x10000) / 0x400 + 0xD800), 403 (int)((uch - 0x10000) / 0x400 + 0xD800),
221 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 404 (int)((uch - 0x10000) % 0x400 + 0xDC00));
222 enc->cur += 12; 405 enc->cur += 12;
223 } 406 }
224 else 407 else
225 { 408 {
226 static char hexdigit [16] = "0123456789abcdef";
227 need (enc, len += 5); 409 need (enc, len += 5);
228 *enc->cur++ = '\\'; 410 *enc->cur++ = '\\';
229 *enc->cur++ = 'u'; 411 *enc->cur++ = 'u';
230 *enc->cur++ = hexdigit [ uch >> 12 ]; 412 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
231 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 413 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
232 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 414 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
233 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 415 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
234 } 416 }
235 417
236 str += clen; 418 str += clen;
237 } 419 }
238 else if (enc->json.flags & F_LATIN1) 420 else if (enc->json.flags & F_LATIN1)
… …
250 while (--clen); 432 while (--clen);
251 } 433 }
252 else 434 else
253 { 435 {
254 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 436 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
255 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 437 enc->cur = encode_utf8 (enc->cur, uch);
256 ++str; 438 ++str;
257 } 439 }
258 } 440 }
259 } 441 }
260 } 442 }
261 443
262 --len; 444 --len;
263 } 445 }
264} 446}
265 447
266inline void 448INLINE void
267encode_indent (enc_t *enc) 449encode_indent (enc_t *enc)
268{ 450{
269 if (enc->json.flags & F_INDENT) 451 if (enc->json.flags & F_INDENT)
270 { 452 {
271 int spaces = enc->indent * INDENT_STEP; 453 int spaces = enc->indent * INDENT_STEP;
… …
274 memset (enc->cur, ' ', spaces); 456 memset (enc->cur, ' ', spaces);
275 enc->cur += spaces; 457 enc->cur += spaces;
276 } 458 }
277} 459}
278 460
279inline void 461INLINE void
280encode_space (enc_t *enc) 462encode_space (enc_t *enc)
281{ 463{
282 need (enc, 1); 464 need (enc, 1);
283 encode_ch (enc, ' '); 465 encode_ch (enc, ' ');
284} 466}
285 467
286inline void 468INLINE void
287encode_nl (enc_t *enc) 469encode_nl (enc_t *enc)
288{ 470{
289 if (enc->json.flags & F_INDENT) 471 if (enc->json.flags & F_INDENT)
290 { 472 {
291 need (enc, 1); 473 need (enc, 1);
292 encode_ch (enc, '\n'); 474 encode_ch (enc, '\n');
293 } 475 }
294} 476}
295 477
296inline void 478INLINE void
297encode_comma (enc_t *enc) 479encode_comma (enc_t *enc)
298{ 480{
299 encode_ch (enc, ','); 481 encode_ch (enc, ',');
300 482
301 if (enc->json.flags & F_INDENT) 483 if (enc->json.flags & F_INDENT)
… …
309static void 491static void
310encode_av (enc_t *enc, AV *av) 492encode_av (enc_t *enc, AV *av)
311{ 493{
312 int i, len = av_len (av); 494 int i, len = av_len (av);
313 495
314 if (enc->indent >= enc->maxdepth) 496 if (enc->indent >= enc->json.max_depth)
315 croak ("data structure too deep (hit recursion limit)"); 497 croak (ERR_NESTING_EXCEEDED);
316 498
317 encode_ch (enc, '['); 499 encode_ch (enc, '[');
318 500
319 if (len >= 0) 501 if (len >= 0)
320 { 502 {
321 encode_nl (enc); ++enc->indent; 503 encode_nl (enc); ++enc->indent;
322 504
323 for (i = 0; i <= len; ++i) 505 for (i = 0; i <= len; ++i)
… …
335 encode_comma (enc); 517 encode_comma (enc);
336 } 518 }
337 519
338 encode_nl (enc); --enc->indent; encode_indent (enc); 520 encode_nl (enc); --enc->indent; encode_indent (enc);
339 } 521 }
340 522
341 encode_ch (enc, ']'); 523 encode_ch (enc, ']');
342} 524}
343 525
344static void 526static void
345encode_hk (enc_t *enc, HE *he) 527encode_hk (enc_t *enc, HE *he)
… …
349 if (HeKLEN (he) == HEf_SVKEY) 531 if (HeKLEN (he) == HEf_SVKEY)
350 { 532 {
351 SV *sv = HeSVKEY (he); 533 SV *sv = HeSVKEY (he);
352 STRLEN len; 534 STRLEN len;
353 char *str; 535 char *str;
354 536
355 SvGETMAGIC (sv); 537 SvGETMAGIC (sv);
356 str = SvPV (sv, len); 538 str = SvPV (sv, len);
357 539
358 encode_str (enc, str, len, SvUTF8 (sv)); 540 encode_str (enc, str, len, SvUTF8 (sv));
359 } 541 }
… …
394 576
395static void 577static void
396encode_hv (enc_t *enc, HV *hv) 578encode_hv (enc_t *enc, HV *hv)
397{ 579{
398 HE *he; 580 HE *he;
399 int count;
400 581
401 if (enc->indent >= enc->maxdepth) 582 if (enc->indent >= enc->json.max_depth)
402 croak ("data structure too deep (hit recursion limit)"); 583 croak (ERR_NESTING_EXCEEDED);
403 584
404 encode_ch (enc, '{'); 585 encode_ch (enc, '{');
405 586
406 // for canonical output we have to sort by keys first 587 // for canonical output we have to sort by keys first
407 // actually, this is mostly due to the stupid so-called 588 // actually, this is mostly due to the stupid so-called
408 // security workaround added somewhere in 5.8.x. 589 // security workaround added somewhere in 5.8.x
409 // that randomises hash orderings 590 // that randomises hash orderings
410 if (enc->json.flags & F_CANONICAL) 591 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
411 { 592 {
412 int count = hv_iterinit (hv); 593 int count = hv_iterinit (hv);
413 594
414 if (SvMAGICAL (hv)) 595 if (SvMAGICAL (hv))
415 { 596 {
… …
426 } 607 }
427 608
428 if (count) 609 if (count)
429 { 610 {
430 int i, fast = 1; 611 int i, fast = 1;
431#if defined(__BORLANDC__) || defined(_MSC_VER) 612 HE *hes_stack [STACK_HES];
432 HE **hes = _alloca (count * sizeof (HE)); 613 HE **hes = hes_stack;
433#else 614
434 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 615 // allocate larger arrays on the heap
435#endif 616 if (count > STACK_HES)
617 {
618 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
619 hes = (HE **)SvPVX (sv);
620 }
436 621
437 i = 0; 622 i = 0;
438 while ((he = hv_iternext (hv))) 623 while ((he = hv_iternext (hv)))
439 { 624 {
440 hes [i++] = he; 625 hes [i++] = he;
… …
509// encode objects, arrays and special \0=false and \1=true values. 694// encode objects, arrays and special \0=false and \1=true values.
510static void 695static void
511encode_rv (enc_t *enc, SV *sv) 696encode_rv (enc_t *enc, SV *sv)
512{ 697{
513 svtype svt; 698 svtype svt;
699 GV *method;
514 700
515 SvGETMAGIC (sv); 701 SvGETMAGIC (sv);
516 svt = SvTYPE (sv); 702 svt = SvTYPE (sv);
517 703
518 if (expect_false (SvOBJECT (sv))) 704 if (expect_false (SvOBJECT (sv)))
519 { 705 {
520 HV *stash = !JSON_SLOW || json_boolean_stash 706 HV *boolean_stash = !JSON_SLOW || types_boolean_stash
521 ? json_boolean_stash 707 ? types_boolean_stash
522 : gv_stashpv ("JSON::XS::Boolean", 1); 708 : gv_stashpv ("Types::Serialiser::Boolean", 1);
709 HV *stash = SvSTASH (sv);
523 710
524 if (SvSTASH (sv) == stash) 711 if (stash == boolean_stash)
525 { 712 {
526 if (SvIV (sv)) 713 if (SvIV (sv))
527 encode_str (enc, "true", 4, 0); 714 encode_str (enc, "true", 4, 0);
528 else 715 else
529 encode_str (enc, "false", 5, 0); 716 encode_str (enc, "false", 5, 0);
530 } 717 }
718 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
719 {
720 int count;
721 dSP;
722
723 ENTER; SAVETMPS; PUSHMARK (SP);
724 EXTEND (SP, 2);
725 // we re-bless the reference to get overload and other niceties right
726 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
727 PUSHs (sv_json);
728
729 PUTBACK;
730 count = call_sv ((SV *)GvCV (method), G_ARRAY);
731 SPAGAIN;
732
733 // catch this surprisingly common error
734 if (SvROK (TOPs) && SvRV (TOPs) == sv)
735 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
736
737 encode_ch (enc, '(');
738 encode_ch (enc, '"');
739 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
740 encode_ch (enc, '"');
741 encode_ch (enc, ')');
742 encode_ch (enc, '[');
743
744 while (count)
745 {
746 encode_sv (enc, SP[1 - count--]);
747
748 if (count)
749 encode_ch (enc, ',');
750 }
751
752 encode_ch (enc, ']');
753
754 PUTBACK;
755
756 FREETMPS; LEAVE;
757 }
758 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
759 {
760 dSP;
761
762 ENTER; SAVETMPS; PUSHMARK (SP);
763 // we re-bless the reference to get overload and other niceties right
764 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
765
766 // calling with G_SCALAR ensures that we always get a 1 return value
767 PUTBACK;
768 call_sv ((SV *)GvCV (method), G_SCALAR);
769 SPAGAIN;
770
771 // catch this surprisingly common error
772 if (SvROK (TOPs) && SvRV (TOPs) == sv)
773 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
774
775 sv = POPs;
776 PUTBACK;
777
778 encode_sv (enc, sv);
779
780 FREETMPS; LEAVE;
781 }
782 else if (enc->json.flags & F_ALLOW_BLESSED)
783 encode_str (enc, "null", 4, 0);
531 else 784 else
532 {
533#if 0
534 if (0 && sv_derived_from (rv, "JSON::Literal"))
535 {
536 // not yet
537 }
538#endif
539 if (enc->json.flags & F_CONV_BLESSED)
540 {
541 // we re-bless the reference to get overload and other niceties right
542 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
543
544 if (to_json)
545 {
546 dSP;
547
548 ENTER; SAVETMPS; PUSHMARK (SP);
549 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
550
551 // calling with G_SCALAR ensures that we always get a 1 return value
552 PUTBACK;
553 call_sv ((SV *)GvCV (to_json), G_SCALAR);
554 SPAGAIN;
555
556 // catch this surprisingly common error
557 if (SvROK (TOPs) && SvRV (TOPs) == sv)
558 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
559
560 sv = POPs;
561 PUTBACK;
562
563 encode_sv (enc, sv);
564
565 FREETMPS; LEAVE;
566 }
567 else if (enc->json.flags & F_ALLOW_BLESSED)
568 encode_str (enc, "null", 4, 0);
569 else
570 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
571 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
572 }
573 else if (enc->json.flags & F_ALLOW_BLESSED)
574 encode_str (enc, "null", 4, 0);
575 else
576 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 785 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
577 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 786 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
578 }
579 } 787 }
580 else if (svt == SVt_PVHV) 788 else if (svt == SVt_PVHV)
581 encode_hv (enc, (HV *)sv); 789 encode_hv (enc, (HV *)sv);
582 else if (svt == SVt_PVAV) 790 else if (svt == SVt_PVAV)
583 encode_av (enc, (AV *)sv); 791 encode_av (enc, (AV *)sv);
… …
588 796
589 if (len == 1 && *pv == '1') 797 if (len == 1 && *pv == '1')
590 encode_str (enc, "true", 4, 0); 798 encode_str (enc, "true", 4, 0);
591 else if (len == 1 && *pv == '0') 799 else if (len == 1 && *pv == '0')
592 encode_str (enc, "false", 5, 0); 800 encode_str (enc, "false", 5, 0);
801 else if (enc->json.flags & F_ALLOW_UNKNOWN)
802 encode_str (enc, "null", 4, 0);
593 else 803 else
594 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 804 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
595 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 805 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
596 } 806 }
807 else if (enc->json.flags & F_ALLOW_UNKNOWN)
808 encode_str (enc, "null", 4, 0);
597 else 809 else
598 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 810 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
599 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 811 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
600} 812}
601 813
… …
619 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 831 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
620 enc->cur += strlen (enc->cur); 832 enc->cur += strlen (enc->cur);
621 } 833 }
622 else if (SvIOKp (sv)) 834 else if (SvIOKp (sv))
623 { 835 {
624 // we assume we can always read an IV as a UV 836 // we assume we can always read an IV as a UV and vice versa
625 if (SvUV (sv) & ~(UV)0x7fff) 837 // we assume two's complement
626 { 838 // we assume no aliasing issues in the union
627 // large integer, use the (rather slow) snprintf way. 839 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
628 need (enc, sizeof (UV) * 3); 840 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
629 enc->cur +=
630 SvIsUV(sv)
631 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
632 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
633 }
634 else
635 { 841 {
636 // optimise the "small number case" 842 // optimise the "small number case"
637 // code will likely be branchless and use only a single multiplication 843 // code will likely be branchless and use only a single multiplication
844 // works for numbers up to 59074
638 I32 i = SvIV (sv); 845 I32 i = SvIVX (sv);
639 U32 u; 846 U32 u;
640 char digit, nz = 0; 847 char digit, nz = 0;
641 848
642 need (enc, 6); 849 need (enc, 6);
643 850
… …
649 856
650 // now output digit by digit, each time masking out the integer part 857 // now output digit by digit, each time masking out the integer part
651 // and multiplying by 5 while moving the decimal point one to the right, 858 // and multiplying by 5 while moving the decimal point one to the right,
652 // resulting in a net multiplication by 10. 859 // resulting in a net multiplication by 10.
653 // we always write the digit to memory but conditionally increment 860 // we always write the digit to memory but conditionally increment
654 // the pointer, to ease the usage of conditional move instructions. 861 // the pointer, to enable the use of conditional move instructions.
655 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 862 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
656 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 863 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
657 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 864 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
658 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 865 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
659 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 866 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
660 } 867 }
868 else
869 {
870 // large integer, use the (rather slow) snprintf way.
871 need (enc, IVUV_MAXCHARS);
872 enc->cur +=
873 SvIsUV(sv)
874 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
875 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
876 }
661 } 877 }
662 else if (SvROK (sv)) 878 else if (SvROK (sv))
663 encode_rv (enc, SvRV (sv)); 879 encode_rv (enc, SvRV (sv));
664 else if (!SvOK (sv)) 880 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
665 encode_str (enc, "null", 4, 0); 881 encode_str (enc, "null", 4, 0);
666 else 882 else
667 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 883 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
668 SvPV_nolen (sv), SvFLAGS (sv)); 884 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
669} 885}
670 886
671static SV * 887static SV *
672encode_json (SV *scalar, JSON *json) 888encode_json (SV *scalar, JSON *json)
673{ 889{
… …
679 enc.json = *json; 895 enc.json = *json;
680 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 896 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
681 enc.cur = SvPVX (enc.sv); 897 enc.cur = SvPVX (enc.sv);
682 enc.end = SvEND (enc.sv); 898 enc.end = SvEND (enc.sv);
683 enc.indent = 0; 899 enc.indent = 0;
684 enc.maxdepth = DEC_DEPTH (enc.json.flags); 900 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
901 : enc.json.flags & F_LATIN1 ? 0x000100UL
902 : 0x110000UL;
685 903
686 SvPOK_only (enc.sv); 904 SvPOK_only (enc.sv);
687 encode_sv (&enc, scalar); 905 encode_sv (&enc, scalar);
906 encode_nl (&enc);
688 907
689 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 908 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
690 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 909 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
691 910
692 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 911 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
710 JSON json; 929 JSON json;
711 U32 depth; // recursion depth 930 U32 depth; // recursion depth
712 U32 maxdepth; // recursion depth limit 931 U32 maxdepth; // recursion depth limit
713} dec_t; 932} dec_t;
714 933
715inline void 934INLINE void
716decode_comment (dec_t *dec) 935decode_comment (dec_t *dec)
717{ 936{
718 // only '#'-style comments allowed a.t.m. 937 // only '#'-style comments allowed a.t.m.
719 938
720 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) 939 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
721 ++dec->cur; 940 ++dec->cur;
722} 941}
723 942
724inline void 943INLINE void
725decode_ws (dec_t *dec) 944decode_ws (dec_t *dec)
726{ 945{
727 for (;;) 946 for (;;)
728 { 947 {
729 char ch = *dec->cur; 948 char ch = *dec->cur;
… …
753 if (*dec->cur != ch) \ 972 if (*dec->cur != ch) \
754 ERR (# ch " expected"); \ 973 ERR (# ch " expected"); \
755 ++dec->cur; \ 974 ++dec->cur; \
756 SE 975 SE
757 976
758#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 977#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
759#define DEC_DEC_DEPTH --dec->depth 978#define DEC_DEC_DEPTH --dec->depth
760 979
761static SV *decode_sv (dec_t *dec); 980static SV *decode_sv (dec_t *dec);
762 981
763static signed char decode_hexdigit[256]; 982static signed char decode_hexdigit[256];
… …
855 1074
856 if (hi >= 0x80) 1075 if (hi >= 0x80)
857 { 1076 {
858 utf8 = 1; 1077 utf8 = 1;
859 1078
860 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1079 cur = encode_utf8 (cur, hi);
861 } 1080 }
862 else 1081 else
863 *cur++ = hi; 1082 *cur++ = hi;
864 } 1083 }
865 break; 1084 break;
… …
867 default: 1086 default:
868 --dec_cur; 1087 --dec_cur;
869 ERR ("illegal backslash escape sequence in string"); 1088 ERR ("illegal backslash escape sequence in string");
870 } 1089 }
871 } 1090 }
872 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1091 else if (expect_true (ch >= 0x20 && ch < 0x80))
873 *cur++ = ch; 1092 *cur++ = ch;
874 else if (ch >= 0x80) 1093 else if (ch >= 0x80)
875 { 1094 {
876 STRLEN clen; 1095 STRLEN clen;
877 UV uch;
878 1096
879 --dec_cur; 1097 --dec_cur;
880 1098
881 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1099 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
882 if (clen == (STRLEN)-1) 1100 if (clen == (STRLEN)-1)
883 ERR ("malformed UTF-8 character in JSON string"); 1101 ERR ("malformed UTF-8 character in JSON string");
884 1102
885 do 1103 do
886 *cur++ = *dec_cur++; 1104 *cur++ = *dec_cur++;
887 while (--clen); 1105 while (--clen);
888 1106
889 utf8 = 1; 1107 utf8 = 1;
890 } 1108 }
1109 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1110 *cur++ = ch;
891 else 1111 else
892 { 1112 {
893 --dec_cur; 1113 --dec_cur;
894 1114
895 if (!ch) 1115 if (!ch)
… …
903 { 1123 {
904 STRLEN len = cur - buf; 1124 STRLEN len = cur - buf;
905 1125
906 if (sv) 1126 if (sv)
907 { 1127 {
908 SvGROW (sv, SvCUR (sv) + len + 1); 1128 STRLEN cur = SvCUR (sv);
1129
1130 if (SvLEN (sv) <= cur + len)
1131 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1132
909 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1133 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
910 SvCUR_set (sv, SvCUR (sv) + len); 1134 SvCUR_set (sv, SvCUR (sv) + len);
911 } 1135 }
912 else 1136 else
913 sv = newSVpvn (buf, len); 1137 sv = newSVpvn (buf, len);
… …
1000 1224
1001 if (!is_nv) 1225 if (!is_nv)
1002 { 1226 {
1003 int len = dec->cur - start; 1227 int len = dec->cur - start;
1004 1228
1005 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1229 // special case the rather common 1..5-digit-int case
1006 if (*start == '-') 1230 if (*start == '-')
1007 switch (len) 1231 switch (len)
1008 { 1232 {
1009 case 2: return newSViv (-( start [1] - '0' * 1)); 1233 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1010 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1234 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1011 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1235 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1012 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1236 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1237 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1013 } 1238 }
1014 else 1239 else
1015 switch (len) 1240 switch (len)
1016 { 1241 {
1017 case 1: return newSViv ( start [0] - '0' * 1); 1242 case 1: return newSViv ( start [0] - '0' * 1);
1018 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1243 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1019 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1244 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1020 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1245 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1246 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1021 } 1247 }
1022 1248
1023 { 1249 {
1024 UV uv; 1250 UV uv;
1025 int numtype = grok_number (start, len, &uv); 1251 int numtype = grok_number (start, len, &uv);
… …
1034 } 1260 }
1035 1261
1036 len -= *start == '-' ? 1 : 0; 1262 len -= *start == '-' ? 1 : 0;
1037 1263
1038 // does not fit into IV or UV, try NV 1264 // does not fit into IV or UV, try NV
1039 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1265 if (len <= NV_DIG)
1040 #if defined (LDBL_DIG)
1041 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1042 #endif
1043 )
1044 // fits into NV without loss of precision 1266 // fits into NV without loss of precision
1045 return newSVnv (Atof (start)); 1267 return newSVnv (json_atof (start));
1046 1268
1047 // everything else fails, convert it to a string 1269 // everything else fails, convert it to a string
1048 return newSVpvn (start, dec->cur - start); 1270 return newSVpvn (start, dec->cur - start);
1049 } 1271 }
1050 1272
1051 // loss of precision here 1273 // loss of precision here
1052 return newSVnv (Atof (start)); 1274 return newSVnv (json_atof (start));
1053 1275
1054fail: 1276fail:
1055 return 0; 1277 return 0;
1056} 1278}
1057 1279
… …
1081 if (*dec->cur == ']') 1303 if (*dec->cur == ']')
1082 { 1304 {
1083 ++dec->cur; 1305 ++dec->cur;
1084 break; 1306 break;
1085 } 1307 }
1086 1308
1087 if (*dec->cur != ',') 1309 if (*dec->cur != ',')
1088 ERR (", or ] expected while parsing array"); 1310 ERR (", or ] expected while parsing array");
1089 1311
1090 ++dec->cur; 1312 ++dec->cur;
1091 1313
… …
1133 char *p = dec->cur; 1355 char *p = dec->cur;
1134 char *e = p + 24; // only try up to 24 bytes 1356 char *e = p + 24; // only try up to 24 bytes
1135 1357
1136 for (;;) 1358 for (;;)
1137 { 1359 {
1138 // the >= 0x80 is true on most architectures 1360 // the >= 0x80 is false on most architectures
1139 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1361 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1140 { 1362 {
1141 // slow path, back up and use decode_str 1363 // slow path, back up and use decode_str
1142 SV *key = decode_str (dec); 1364 SV *key = decode_str (dec);
1143 if (!key) 1365 if (!key)
… …
1226 dSP; 1448 dSP;
1227 int count; 1449 int count;
1228 1450
1229 ENTER; SAVETMPS; PUSHMARK (SP); 1451 ENTER; SAVETMPS; PUSHMARK (SP);
1230 XPUSHs (HeVAL (he)); 1452 XPUSHs (HeVAL (he));
1453 sv_2mortal (sv);
1231 1454
1232 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1455 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1233 1456
1234 if (count == 1) 1457 if (count == 1)
1235 { 1458 {
1236 sv = newSVsv (POPs); 1459 sv = newSVsv (POPs);
1237 FREETMPS; LEAVE; 1460 FREETMPS; LEAVE;
1238 return sv; 1461 return sv;
1239 } 1462 }
1240 1463
1464 SvREFCNT_inc (sv);
1241 FREETMPS; LEAVE; 1465 FREETMPS; LEAVE;
1242 } 1466 }
1243 } 1467 }
1244 1468
1245 if (dec->json.cb_object) 1469 if (dec->json.cb_object)
… …
1271 DEC_DEC_DEPTH; 1495 DEC_DEC_DEPTH;
1272 return 0; 1496 return 0;
1273} 1497}
1274 1498
1275static SV * 1499static SV *
1500decode_tag (dec_t *dec)
1501{
1502 SV *tag = 0;
1503 SV *val = 0;
1504
1505 if (!(dec->json.flags & F_ALLOW_TAGS))
1506 ERR ("malformed JSON string, neither array, object, number, string or atom");
1507
1508 ++dec->cur;
1509
1510 decode_ws (dec);
1511
1512 tag = decode_sv (dec);
1513 if (!tag)
1514 goto fail;
1515
1516 if (!SvPOK (tag))
1517 ERR ("malformed JSON string, (tag) must be a string");
1518
1519 decode_ws (dec);
1520
1521 if (*dec->cur != ')')
1522 ERR (") expected after tag");
1523
1524 ++dec->cur;
1525
1526 decode_ws (dec);
1527
1528 val = decode_sv (dec);
1529 if (!val)
1530 goto fail;
1531
1532 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1533 ERR ("malformed JSON string, tag value must be an array");
1534
1535 {
1536 AV *av = (AV *)SvRV (val);
1537 int i, len = av_len (av) + 1;
1538 HV *stash = gv_stashsv (tag, 0);
1539 SV *sv;
1540
1541 if (!stash)
1542 ERR ("cannot decode perl-object (package does not exist)");
1543
1544 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1545
1546 if (!method)
1547 ERR ("cannot decode perl-object (package does not have a THAW method)");
1548
1549 dSP;
1550
1551 ENTER; SAVETMPS; PUSHMARK (SP);
1552 EXTEND (SP, len + 2);
1553 // we re-bless the reference to get overload and other niceties right
1554 PUSHs (tag);
1555 PUSHs (sv_json);
1556
1557 for (i = 0; i < len; ++i)
1558 PUSHs (*av_fetch (av, i, 1));
1559
1560 PUTBACK;
1561 call_sv ((SV *)GvCV (method), G_SCALAR);
1562 SPAGAIN;
1563
1564 SvREFCNT_dec (tag);
1565 SvREFCNT_dec (val);
1566 sv = SvREFCNT_inc (POPs);
1567
1568 PUTBACK;
1569
1570 FREETMPS; LEAVE;
1571
1572 return sv;
1573 }
1574
1575fail:
1576 SvREFCNT_dec (tag);
1577 SvREFCNT_dec (val);
1578 return 0;
1579}
1580
1581static SV *
1276decode_sv (dec_t *dec) 1582decode_sv (dec_t *dec)
1277{ 1583{
1278 // the beauty of JSON: you need exactly one character lookahead 1584 // the beauty of JSON: you need exactly one character lookahead
1279 // to parse anything. 1585 // to parse everything.
1280 switch (*dec->cur) 1586 switch (*dec->cur)
1281 { 1587 {
1282 case '"': ++dec->cur; return decode_str (dec); 1588 case '"': ++dec->cur; return decode_str (dec);
1283 case '[': ++dec->cur; return decode_av (dec); 1589 case '[': ++dec->cur; return decode_av (dec);
1284 case '{': ++dec->cur; return decode_hv (dec); 1590 case '{': ++dec->cur; return decode_hv (dec);
1591 case '(': return decode_tag (dec);
1285 1592
1286 case '-': 1593 case '-':
1287 case '0': case '1': case '2': case '3': case '4': 1594 case '0': case '1': case '2': case '3': case '4':
1288 case '5': case '6': case '7': case '8': case '9': 1595 case '5': case '6': case '7': case '8': case '9':
1289 return decode_num (dec); 1596 return decode_num (dec);
… …
1291 case 't': 1598 case 't':
1292 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1599 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1293 { 1600 {
1294 dec->cur += 4; 1601 dec->cur += 4;
1295#if JSON_SLOW 1602#if JSON_SLOW
1296 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1603 types_true = get_bool ("Types::Serialiser::true");
1297#endif 1604#endif
1298 return SvREFCNT_inc (json_true); 1605 return newSVsv (types_true);
1299 } 1606 }
1300 else 1607 else
1301 ERR ("'true' expected"); 1608 ERR ("'true' expected");
1302 1609
1303 break; 1610 break;
… …
1305 case 'f': 1612 case 'f':
1306 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1613 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1307 { 1614 {
1308 dec->cur += 5; 1615 dec->cur += 5;
1309#if JSON_SLOW 1616#if JSON_SLOW
1310 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1617 types_false = get_bool ("Types::Serialiser::false");
1311#endif 1618#endif
1312 return SvREFCNT_inc (json_false); 1619 return newSVsv (types_false);
1313 } 1620 }
1314 else 1621 else
1315 ERR ("'false' expected"); 1622 ERR ("'false' expected");
1316 1623
1317 break; 1624 break;
… …
1326 ERR ("'null' expected"); 1633 ERR ("'null' expected");
1327 1634
1328 break; 1635 break;
1329 1636
1330 default: 1637 default:
1331 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1638 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1332 break; 1639 break;
1333 } 1640 }
1334 1641
1335fail: 1642fail:
1336 return 0; 1643 return 0;
1337} 1644}
1338 1645
1339static SV * 1646static SV *
1340decode_json (SV *string, JSON *json, UV *offset_return) 1647decode_json (SV *string, JSON *json, char **offset_return)
1341{ 1648{
1342 dec_t dec; 1649 dec_t dec;
1343 UV offset;
1344 SV *sv; 1650 SV *sv;
1345 1651
1652 /* work around bugs in 5.10 where manipulating magic values
1653 * makes perl ignore the magic in subsequent accesses.
1654 * also make a copy of non-PV values, to get them into a clean
1655 * state (SvPV should do that, but it's buggy, see below).
1656 */
1346 SvGETMAGIC (string); 1657 /*SvGETMAGIC (string);*/
1658 if (SvMAGICAL (string) || !SvPOK (string))
1659 string = sv_2mortal (newSVsv (string));
1660
1347 SvUPGRADE (string, SVt_PV); 1661 SvUPGRADE (string, SVt_PV);
1348 1662
1349 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1663 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1664 * assertion with -DDEBUGGING, although SvCUR is documented to
1665 * return the xpv_cur field which certainly exists after upgrading.
1666 * according to nicholas clark, calling SvPOK fixes this.
1667 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1668 * and hope for the best.
1669 * Damnit, SvPV_nolen still trips over yet another assertion. This
1670 * assertion business is seriously broken, try yet another workaround
1671 * for the broken -DDEBUGGING.
1672 */
1673 {
1674#ifdef DEBUGGING
1675 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1676#else
1677 STRLEN offset = SvCUR (string);
1678#endif
1679
1680 if (offset > json->max_size && json->max_size)
1350 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1681 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1351 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1682 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1683 }
1352 1684
1353 if (json->flags & F_UTF8) 1685 if (DECODE_WANTS_OCTETS (json))
1354 sv_utf8_downgrade (string, 0); 1686 sv_utf8_downgrade (string, 0);
1355 else 1687 else
1356 sv_utf8_upgrade (string); 1688 sv_utf8_upgrade (string);
1357 1689
1358 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1690 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1359 1691
1360 dec.json = *json; 1692 dec.json = *json;
1361 dec.cur = SvPVX (string); 1693 dec.cur = SvPVX (string);
1362 dec.end = SvEND (string); 1694 dec.end = SvEND (string);
1363 dec.err = 0; 1695 dec.err = 0;
1364 dec.depth = 0; 1696 dec.depth = 0;
1365 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1366 1697
1367 if (dec.json.cb_object || dec.json.cb_sk_object) 1698 if (dec.json.cb_object || dec.json.cb_sk_object)
1368 dec.json.flags |= F_HOOK; 1699 dec.json.flags |= F_HOOK;
1369 1700
1370 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1701 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1371 1702
1372 decode_ws (&dec); 1703 decode_ws (&dec);
1373 sv = decode_sv (&dec); 1704 sv = decode_sv (&dec);
1705
1706 if (offset_return)
1707 *offset_return = dec.cur;
1374 1708
1375 if (!(offset_return || !sv)) 1709 if (!(offset_return || !sv))
1376 { 1710 {
1377 // check for trailing garbage 1711 // check for trailing garbage
1378 decode_ws (&dec); 1712 decode_ws (&dec);
… …
1381 { 1715 {
1382 dec.err = "garbage after JSON object"; 1716 dec.err = "garbage after JSON object";
1383 SvREFCNT_dec (sv); 1717 SvREFCNT_dec (sv);
1384 sv = 0; 1718 sv = 0;
1385 } 1719 }
1386 }
1387
1388 if (offset_return || !sv)
1389 {
1390 offset = dec.json.flags & F_UTF8
1391 ? dec.cur - SvPVX (string)
1392 : utf8_distance (dec.cur, SvPVX (string));
1393
1394 if (offset_return)
1395 *offset_return = offset;
1396 } 1720 }
1397 1721
1398 if (!sv) 1722 if (!sv)
1399 { 1723 {
1400 SV *uni = sv_newmortal (); 1724 SV *uni = sv_newmortal ();
… …
1406 SAVEVPTR (PL_curcop); 1730 SAVEVPTR (PL_curcop);
1407 PL_curcop = &cop; 1731 PL_curcop = &cop;
1408 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1732 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1409 LEAVE; 1733 LEAVE;
1410 1734
1411 croak ("%s, at character offset %d [\"%s\"]", 1735 croak ("%s, at character offset %d (before \"%s\")",
1412 dec.err, 1736 dec.err,
1413 (int)offset, 1737 (int)ptr_to_index (string, dec.cur),
1414 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1738 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1415 } 1739 }
1416 1740
1417 sv = sv_2mortal (sv); 1741 sv = sv_2mortal (sv);
1418 1742
1419 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1743 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1420 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1744 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1421 1745
1422 return sv; 1746 return sv;
1747}
1748
1749/////////////////////////////////////////////////////////////////////////////
1750// incremental parser
1751
1752static void
1753incr_parse (JSON *self)
1754{
1755 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1756
1757 // the state machine here is a bit convoluted and could be simplified a lot
1758 // but this would make it slower, so...
1759
1760 for (;;)
1761 {
1762 //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
1763 switch (self->incr_mode)
1764 {
1765 // only used for initial whitespace skipping
1766 case INCR_M_WS:
1767 for (;;)
1768 {
1769 if (*p > 0x20)
1770 {
1771 if (*p == '#')
1772 {
1773 self->incr_mode = INCR_M_C0;
1774 goto incr_m_c;
1775 }
1776 else
1777 {
1778 self->incr_mode = INCR_M_JSON;
1779 goto incr_m_json;
1780 }
1781 }
1782 else if (!*p)
1783 goto interrupt;
1784
1785 ++p;
1786 }
1787
1788 // skip a single char inside a string (for \\-processing)
1789 case INCR_M_BS:
1790 if (!*p)
1791 goto interrupt;
1792
1793 ++p;
1794 self->incr_mode = INCR_M_STR;
1795 goto incr_m_str;
1796
1797 // inside #-style comments
1798 case INCR_M_C0:
1799 case INCR_M_C1:
1800 incr_m_c:
1801 for (;;)
1802 {
1803 if (*p == '\n')
1804 {
1805 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1806 break;
1807 }
1808 else if (!*p)
1809 goto interrupt;
1810
1811 ++p;
1812 }
1813
1814 break;
1815
1816 // inside a string
1817 case INCR_M_STR:
1818 incr_m_str:
1819 for (;;)
1820 {
1821 if (*p == '"')
1822 {
1823 ++p;
1824 self->incr_mode = INCR_M_JSON;
1825
1826 if (!self->incr_nest)
1827 goto interrupt;
1828
1829 goto incr_m_json;
1830 }
1831 else if (*p == '\\')
1832 {
1833 ++p; // "virtually" consumes character after \
1834
1835 if (!*p) // if at end of string we have to switch modes
1836 {
1837 self->incr_mode = INCR_M_BS;
1838 goto interrupt;
1839 }
1840 }
1841 else if (!*p)
1842 goto interrupt;
1843
1844 ++p;
1845 }
1846
1847 // after initial ws, outside string
1848 case INCR_M_JSON:
1849 incr_m_json:
1850 for (;;)
1851 {
1852 switch (*p++)
1853 {
1854 case 0:
1855 --p;
1856 goto interrupt;
1857
1858 case 0x09:
1859 case 0x0a:
1860 case 0x0d:
1861 case 0x20:
1862 if (!self->incr_nest)
1863 {
1864 --p; // do not eat the whitespace, let the next round do it
1865 goto interrupt;
1866 }
1867 break;
1868
1869 case '"':
1870 self->incr_mode = INCR_M_STR;
1871 goto incr_m_str;
1872
1873 case '[':
1874 case '{':
1875 case '(':
1876 if (++self->incr_nest > self->max_depth)
1877 croak (ERR_NESTING_EXCEEDED);
1878 break;
1879
1880 case ']':
1881 case '}':
1882 if (--self->incr_nest <= 0)
1883 goto interrupt;
1884 break;
1885
1886 case ')':
1887 --self->incr_nest;
1888 break;
1889
1890 case '#':
1891 self->incr_mode = INCR_M_C1;
1892 goto incr_m_c;
1893 }
1894 }
1895 }
1896
1897 modechange:
1898 ;
1899 }
1900
1901interrupt:
1902 self->incr_pos = p - SvPVX (self->incr_text);
1903 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1904 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1423} 1905}
1424 1906
1425///////////////////////////////////////////////////////////////////////////// 1907/////////////////////////////////////////////////////////////////////////////
1426// XS interface functions 1908// XS interface functions
1427 1909
… …
1436 i >= '0' && i <= '9' ? i - '0' 1918 i >= '0' && i <= '9' ? i - '0'
1437 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1919 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1438 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1920 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1439 : -1; 1921 : -1;
1440 1922
1441 json_stash = gv_stashpv ("JSON::XS" , 1); 1923 json_stash = gv_stashpv ("JSON::XS" , 1);
1442 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1924 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1443 1925
1444 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1926 types_true = get_bool ("Types::Serialiser::true");
1445 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1927 types_false = get_bool ("Types::Serialiser::false");
1928
1929 sv_json = newSVpv ("JSON", 0);
1930 SvREADONLY_on (sv_json);
1931
1932 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1446} 1933}
1447 1934
1448PROTOTYPES: DISABLE 1935PROTOTYPES: DISABLE
1449 1936
1450void CLONE (...) 1937void CLONE (...)
1451 CODE: 1938 CODE:
1452 json_stash = 0; 1939 json_stash = 0;
1453 json_boolean_stash = 0; 1940 types_boolean_stash = 0;
1454 1941
1455void new (char *klass) 1942void new (char *klass)
1456 PPCODE: 1943 PPCODE:
1457{ 1944{
1458 SV *pv = NEWSV (0, sizeof (JSON)); 1945 SV *pv = NEWSV (0, sizeof (JSON));
1459 SvPOK_only (pv); 1946 SvPOK_only (pv);
1460 Zero (SvPVX (pv), 1, JSON); 1947 json_init ((JSON *)SvPVX (pv));
1461 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1948 XPUSHs (sv_2mortal (sv_bless (
1462 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH))); 1949 newRV_noinc (pv),
1950 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1951 )));
1463} 1952}
1464 1953
1465void ascii (JSON *self, int enable = 1) 1954void ascii (JSON *self, int enable = 1)
1466 ALIAS: 1955 ALIAS:
1467 ascii = F_ASCII 1956 ascii = F_ASCII
… …
1475 allow_nonref = F_ALLOW_NONREF 1964 allow_nonref = F_ALLOW_NONREF
1476 shrink = F_SHRINK 1965 shrink = F_SHRINK
1477 allow_blessed = F_ALLOW_BLESSED 1966 allow_blessed = F_ALLOW_BLESSED
1478 convert_blessed = F_CONV_BLESSED 1967 convert_blessed = F_CONV_BLESSED
1479 relaxed = F_RELAXED 1968 relaxed = F_RELAXED
1969 allow_unknown = F_ALLOW_UNKNOWN
1970 allow_tags = F_ALLOW_TAGS
1480 PPCODE: 1971 PPCODE:
1481{ 1972{
1482 if (enable) 1973 if (enable)
1483 self->flags |= ix; 1974 self->flags |= ix;
1484 else 1975 else
… …
1494 get_utf8 = F_UTF8 1985 get_utf8 = F_UTF8
1495 get_indent = F_INDENT 1986 get_indent = F_INDENT
1496 get_canonical = F_CANONICAL 1987 get_canonical = F_CANONICAL
1497 get_space_before = F_SPACE_BEFORE 1988 get_space_before = F_SPACE_BEFORE
1498 get_space_after = F_SPACE_AFTER 1989 get_space_after = F_SPACE_AFTER
1499 get_pretty = F_PRETTY
1500 get_allow_nonref = F_ALLOW_NONREF 1990 get_allow_nonref = F_ALLOW_NONREF
1501 get_shrink = F_SHRINK 1991 get_shrink = F_SHRINK
1502 get_allow_blessed = F_ALLOW_BLESSED 1992 get_allow_blessed = F_ALLOW_BLESSED
1503 get_convert_blessed = F_CONV_BLESSED 1993 get_convert_blessed = F_CONV_BLESSED
1504 get_relaxed = F_RELAXED 1994 get_relaxed = F_RELAXED
1995 get_allow_unknown = F_ALLOW_UNKNOWN
1996 get_allow_tags = F_ALLOW_TAGS
1505 PPCODE: 1997 PPCODE:
1506 XPUSHs (boolSV (self->flags & ix)); 1998 XPUSHs (boolSV (self->flags & ix));
1507 1999
1508void max_depth (JSON *self, UV max_depth = 0x80000000UL) 2000void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1509 PPCODE: 2001 PPCODE:
1510{ 2002 self->max_depth = max_depth;
1511 UV log2 = 0;
1512
1513 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1514
1515 while ((1UL << log2) < max_depth)
1516 ++log2;
1517
1518 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1519
1520 XPUSHs (ST (0)); 2003 XPUSHs (ST (0));
1521}
1522 2004
1523int get_max_depth (JSON *self) 2005U32 get_max_depth (JSON *self)
1524 CODE: 2006 CODE:
1525 RETVAL = DEC_DEPTH (self->flags); 2007 RETVAL = self->max_depth;
1526 OUTPUT: 2008 OUTPUT:
1527 RETVAL 2009 RETVAL
1528 2010
1529void max_size (JSON *self, UV max_size = 0) 2011void max_size (JSON *self, U32 max_size = 0)
1530 PPCODE: 2012 PPCODE:
1531{ 2013 self->max_size = max_size;
1532 UV log2 = 0;
1533
1534 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1535 if (max_size == 1) max_size = 2;
1536
1537 while ((1UL << log2) < max_size)
1538 ++log2;
1539
1540 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1541
1542 XPUSHs (ST (0)); 2014 XPUSHs (ST (0));
1543}
1544 2015
1545int get_max_size (JSON *self) 2016int get_max_size (JSON *self)
1546 CODE: 2017 CODE:
1547 RETVAL = DEC_SIZE (self->flags); 2018 RETVAL = self->max_size;
1548 OUTPUT: 2019 OUTPUT:
1549 RETVAL 2020 RETVAL
1550 2021
1551void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 2022void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1552 PPCODE: 2023 PPCODE:
… …
1558} 2029}
1559 2030
1560void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2031void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1561 PPCODE: 2032 PPCODE:
1562{ 2033{
1563 if (!self->cb_sk_object) 2034 if (!self->cb_sk_object)
1564 self->cb_sk_object = newHV (); 2035 self->cb_sk_object = newHV ();
1565 2036
1566 if (SvOK (cb)) 2037 if (SvOK (cb))
1567 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2038 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1568 else 2039 else
… …
1579 XPUSHs (ST (0)); 2050 XPUSHs (ST (0));
1580} 2051}
1581 2052
1582void encode (JSON *self, SV *scalar) 2053void encode (JSON *self, SV *scalar)
1583 PPCODE: 2054 PPCODE:
1584 XPUSHs (encode_json (scalar, self)); 2055 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2056 XPUSHs (scalar);
1585 2057
1586void decode (JSON *self, SV *jsonstr) 2058void decode (JSON *self, SV *jsonstr)
1587 PPCODE: 2059 PPCODE:
1588 XPUSHs (decode_json (jsonstr, self, 0)); 2060 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2061 XPUSHs (jsonstr);
1589 2062
1590void decode_prefix (JSON *self, SV *jsonstr) 2063void decode_prefix (JSON *self, SV *jsonstr)
1591 PPCODE: 2064 PPCODE:
1592{ 2065{
1593 UV offset; 2066 SV *sv;
2067 char *offset;
2068 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1594 EXTEND (SP, 2); 2069 EXTEND (SP, 2);
1595 PUSHs (decode_json (jsonstr, self, &offset)); 2070 PUSHs (sv);
1596 PUSHs (sv_2mortal (newSVuv (offset))); 2071 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
2072}
2073
2074void incr_parse (JSON *self, SV *jsonstr = 0)
2075 PPCODE:
2076{
2077 if (!self->incr_text)
2078 self->incr_text = newSVpvn ("", 0);
2079
2080 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2081 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2082 if (DECODE_WANTS_OCTETS (self))
2083 {
2084 if (self->incr_pos)
2085 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2086 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2087
2088 sv_utf8_downgrade (self->incr_text, 0);
2089 }
2090 else
2091 {
2092 sv_utf8_upgrade (self->incr_text);
2093
2094 if (self->incr_pos)
2095 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2096 - (U8 *)SvPVX (self->incr_text);
2097 }
2098
2099 // append data, if any
2100 if (jsonstr)
2101 {
2102 /* make sure both strings have same encoding */
2103 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2104 if (SvUTF8 (jsonstr))
2105 sv_utf8_downgrade (jsonstr, 0);
2106 else
2107 sv_utf8_upgrade (jsonstr);
2108
2109 /* and then just blindly append */
2110 {
2111 STRLEN len;
2112 const char *str = SvPV (jsonstr, len);
2113 STRLEN cur = SvCUR (self->incr_text);
2114
2115 if (SvLEN (self->incr_text) <= cur + len)
2116 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
2117
2118 Move (str, SvEND (self->incr_text), len, char);
2119 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2120 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2121 }
2122 }
2123
2124 if (GIMME_V != G_VOID)
2125 do
2126 {
2127 SV *sv;
2128 char *offset;
2129
2130 if (!INCR_DONE (self))
2131 {
2132 incr_parse (self);
2133
2134 if (self->incr_pos > self->max_size && self->max_size)
2135 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2136 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2137
2138 if (!INCR_DONE (self))
2139 {
2140 // as an optimisation, do not accumulate white space in the incr buffer
2141 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2142 {
2143 self->incr_pos = 0;
2144 SvCUR_set (self->incr_text, 0);
2145 }
2146
2147 break;
2148 }
2149 }
2150
2151 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2152 XPUSHs (sv);
2153
2154 self->incr_pos -= offset - SvPVX (self->incr_text);
2155 self->incr_nest = 0;
2156 self->incr_mode = 0;
2157
2158 sv_chop (self->incr_text, offset);
2159 }
2160 while (GIMME_V == G_ARRAY);
2161}
2162
2163SV *incr_text (JSON *self)
2164 ATTRS: lvalue
2165 CODE:
2166{
2167 if (self->incr_pos)
2168 croak ("incr_text can not be called when the incremental parser already started parsing");
2169
2170 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2171}
2172 OUTPUT:
2173 RETVAL
2174
2175void incr_skip (JSON *self)
2176 CODE:
2177{
2178 if (self->incr_pos)
2179 {
2180 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2181 self->incr_pos = 0;
2182 self->incr_nest = 0;
2183 self->incr_mode = 0;
2184 }
2185}
2186
2187void incr_reset (JSON *self)
2188 CODE:
2189{
2190 SvREFCNT_dec (self->incr_text);
2191 self->incr_text = 0;
2192 self->incr_pos = 0;
2193 self->incr_nest = 0;
2194 self->incr_mode = 0;
1597} 2195}
1598 2196
1599void DESTROY (JSON *self) 2197void DESTROY (JSON *self)
1600 CODE: 2198 CODE:
1601 SvREFCNT_dec (self->cb_sk_object); 2199 SvREFCNT_dec (self->cb_sk_object);
1602 SvREFCNT_dec (self->cb_object); 2200 SvREFCNT_dec (self->cb_object);
2201 SvREFCNT_dec (self->incr_text);
1603 2202
1604PROTOTYPES: ENABLE 2203PROTOTYPES: ENABLE
1605 2204
1606void to_json (SV *scalar) 2205void encode_json (SV *scalar)
1607 PPCODE: 2206 PPCODE:
1608{ 2207{
1609 JSON json = { F_DEFAULT | F_UTF8 }; 2208 JSON json;
1610 XPUSHs (encode_json (scalar, &json)); 2209 json_init (&json);
2210 json.flags |= F_UTF8;
2211 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2212 XPUSHs (scalar);
1611} 2213}
1612 2214
1613void from_json (SV *jsonstr) 2215void decode_json (SV *jsonstr)
1614 PPCODE: 2216 PPCODE:
1615{ 2217{
1616 JSON json = { F_DEFAULT | F_UTF8 }; 2218 JSON json;
1617 XPUSHs (decode_json (jsonstr, &json, 0)); 2219 json_init (&json);
2220 json.flags |= F_UTF8;
2221 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2222 XPUSHs (jsonstr);
1618} 2223}
1619 2224

Diff Legend

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