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 1.48 by root, Sun Jul 1 22:20:00 2007 UTC vs.
Revision 1.119 by root, Mon Oct 28 23:19:54 2013 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include "stdio.h" 8#include <stdio.h>
9#include <limits.h>
10#include <float.h>
9 11
10#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 14#endif
13 15
14// some old perls do not have this, try to make it work, no 16// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces. 17// guarantees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
18#endif 20#endif
21
22// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
19 24
20#define F_ASCII 0x00000001UL 25#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL 26#define F_LATIN1 0x00000002UL
22#define F_UTF8 0x00000004UL 27#define F_UTF8 0x00000004UL
23#define F_INDENT 0x00000008UL 28#define F_INDENT 0x00000008UL
… …
25#define F_SPACE_BEFORE 0x00000020UL 30#define F_SPACE_BEFORE 0x00000020UL
26#define F_SPACE_AFTER 0x00000040UL 31#define F_SPACE_AFTER 0x00000040UL
27#define F_ALLOW_NONREF 0x00000100UL 32#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL 33#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL 34#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL // NYI 35#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL 36#define F_RELAXED 0x00001000UL
34#define S_MAXSIZE 20 37#define F_ALLOW_UNKNOWN 0x00002000UL
35 38#define F_ALLOW_TAGS 0x00004000UL
36#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 39#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
37#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
38 40
39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
40#define F_DEFAULT (9UL << S_MAXDEPTH)
41 42
42#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
43#define INDENT_STEP 3 // spaces per indentation level 44#define INDENT_STEP 3 // spaces per indentation level
44 45
45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
46 47
48#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
49
47#define SB do { 50#define SB do {
48#define SE } while (0) 51#define SE } while (0)
49 52
50#if __GNUC__ >= 3 53#if __GNUC__ >= 3
51# define expect(expr,value) __builtin_expect ((expr),(value)) 54# define expect(expr,value) __builtin_expect ((expr), (value))
52# define inline inline 55# define INLINE static inline
53#else 56#else
54# define expect(expr,value) (expr) 57# define expect(expr,value) (expr)
55# define inline static 58# define INLINE static
56#endif 59#endif
57 60
58#define expect_false(expr) expect ((expr) != 0, 0) 61#define expect_false(expr) expect ((expr) != 0, 0)
59#define expect_true(expr) expect ((expr) != 0, 1) 62#define expect_true(expr) expect ((expr) != 0, 1)
60 63
64#define IN_RANGE_INC(type,val,beg,end) \
65 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
66 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
67
68#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
69
70#ifdef USE_ITHREADS
71# define JSON_SLOW 1
72# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
73#else
74# define JSON_SLOW 0
75# define JSON_STASH json_stash
76#endif
77
78// the amount of HEs to allocate on the stack, when sorting keys
79#define STACK_HES 64
80
61static HV *json_stash, *json_boolean_stash; // JSON::XS:: 81static HV *json_stash, *types_boolean_stash; // JSON::XS::
62static SV *json_true, *json_false; 82static SV *types_true, *types_false, *sv_json;
83
84enum {
85 INCR_M_WS = 0, // initial whitespace skipping, must be 0
86 INCR_M_STR, // inside string
87 INCR_M_BS, // inside backslash
88 INCR_M_C0, // inside comment in initial whitespace sequence
89 INCR_M_C1, // inside comment in other places
90 INCR_M_JSON // outside anything, count nesting
91};
92
93#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
63 94
64typedef struct { 95typedef struct {
65 U32 flags; 96 U32 flags;
97 U32 max_depth;
98 STRLEN max_size;
99
100 SV *cb_object;
101 HV *cb_sk_object;
102
103 // for the incremental parser
104 SV *incr_text; // the source text so far
105 STRLEN incr_pos; // the current offset into the text
106 int incr_nest; // {[]}-nesting level
107 unsigned char incr_mode;
66} JSON; 108} JSON;
109
110INLINE void
111json_init (JSON *json)
112{
113 Zero (json, 1, JSON);
114 json->max_depth = 512;
115}
67 116
68///////////////////////////////////////////////////////////////////////////// 117/////////////////////////////////////////////////////////////////////////////
69// utility functions 118// utility functions
70 119
71inline void 120INLINE SV *
121get_bool (const char *name)
122{
123 SV *sv = get_sv (name, 1);
124
125 SvREADONLY_on (sv);
126 SvREADONLY_on (SvRV (sv));
127
128 return sv;
129}
130
131INLINE void
72shrink (SV *sv) 132shrink (SV *sv)
73{ 133{
74 sv_utf8_downgrade (sv, 1); 134 sv_utf8_downgrade (sv, 1);
135
75 if (SvLEN (sv) > SvCUR (sv) + 1) 136 if (SvLEN (sv) > SvCUR (sv) + 1)
76 { 137 {
77#ifdef SvPV_shrink_to_cur 138#ifdef SvPV_shrink_to_cur
78 SvPV_shrink_to_cur (sv); 139 SvPV_shrink_to_cur (sv);
79#elif defined (SvPV_renew) 140#elif defined (SvPV_renew)
… …
85// decode an utf-8 character and return it, or (UV)-1 in 146// decode an utf-8 character and return it, or (UV)-1 in
86// case of an error. 147// case of an error.
87// we special-case "safe" characters from U+80 .. U+7FF, 148// we special-case "safe" characters from U+80 .. U+7FF,
88// but use the very good perl function to parse anything else. 149// but use the very good perl function to parse anything else.
89// note that we never call this function for a ascii codepoints 150// note that we never call this function for a ascii codepoints
90inline UV 151INLINE UV
91decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 152decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
92{ 153{
93 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 154 if (expect_true (len >= 2
94 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 155 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
95 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 156 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
96 { 157 {
97 *clen = 2; 158 *clen = 2;
98 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 159 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
99 } 160 }
100 else 161 else
101 { 162 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
102 *clen = (STRLEN)-1; 163}
103 return (UV)-1; 164
165// likewise for encoding, also never called for ascii codepoints
166// this function takes advantage of this fact, although current gccs
167// seem to optimise the check for >= 0x80 away anyways
168INLINE unsigned char *
169encode_utf8 (unsigned char *s, UV ch)
170{
171 if (expect_false (ch < 0x000080))
172 *s++ = ch;
173 else if (expect_true (ch < 0x000800))
174 *s++ = 0xc0 | ( ch >> 6),
175 *s++ = 0x80 | ( ch & 0x3f);
176 else if ( ch < 0x010000)
177 *s++ = 0xe0 | ( ch >> 12),
178 *s++ = 0x80 | ((ch >> 6) & 0x3f),
179 *s++ = 0x80 | ( ch & 0x3f);
180 else if ( ch < 0x110000)
181 *s++ = 0xf0 | ( ch >> 18),
182 *s++ = 0x80 | ((ch >> 12) & 0x3f),
183 *s++ = 0x80 | ((ch >> 6) & 0x3f),
184 *s++ = 0x80 | ( ch & 0x3f);
185
186 return s;
187}
188
189// convert offset pointer to character index, sv must be string
190static STRLEN
191ptr_to_index (SV *sv, char *offset)
192{
193 return SvUTF8 (sv)
194 ? utf8_distance (offset, SvPVX (sv))
195 : offset - SvPVX (sv);
196}
197
198/////////////////////////////////////////////////////////////////////////////
199// fp hell
200
201// scan a group of digits, and a trailing exponent
202static void
203json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
204{
205 UV uaccum = 0;
206 int eaccum = 0;
207
208 // if we recurse too deep, skip all remaining digits
209 // to avoid a stack overflow attack
210 if (expect_false (--maxdepth <= 0))
211 while (((U8)*s - '0') < 10)
212 ++s;
213
214 for (;;)
104 } 215 {
105} 216 U8 dig = (U8)*s - '0';
106 217
218 if (expect_false (dig >= 10))
219 {
220 if (dig == (U8)((U8)'.' - (U8)'0'))
221 {
222 ++s;
223 json_atof_scan1 (s, accum, expo, 1, maxdepth);
224 }
225 else if ((dig | ' ') == 'e' - '0')
226 {
227 int exp2 = 0;
228 int neg = 0;
229
230 ++s;
231
232 if (*s == '-')
233 {
234 ++s;
235 neg = 1;
236 }
237 else if (*s == '+')
238 ++s;
239
240 while ((dig = (U8)*s - '0') < 10)
241 exp2 = exp2 * 10 + *s++ - '0';
242
243 *expo += neg ? -exp2 : exp2;
244 }
245
246 break;
247 }
248
249 ++s;
250
251 uaccum = uaccum * 10 + dig;
252 ++eaccum;
253
254 // if we have too many digits, then recurse for more
255 // we actually do this for rather few digits
256 if (uaccum >= (UV_MAX - 9) / 10)
257 {
258 if (postdp) *expo -= eaccum;
259 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
260 if (postdp) *expo += eaccum;
261
262 break;
263 }
264 }
265
266 // this relies greatly on the quality of the pow ()
267 // implementation of the platform, but a good
268 // implementation is hard to beat.
269 // (IEEE 754 conformant ones are required to be exact)
270 if (postdp) *expo -= eaccum;
271 *accum += uaccum * Perl_pow (10., *expo);
272 *expo += eaccum;
273}
274
275static NV
276json_atof (const char *s)
277{
278 NV accum = 0.;
279 int expo = 0;
280 int neg = 0;
281
282 if (*s == '-')
283 {
284 ++s;
285 neg = 1;
286 }
287
288 // a recursion depth of ten gives us >>500 bits
289 json_atof_scan1 (s, &accum, &expo, 0, 10);
290
291 return neg ? -accum : accum;
292}
107///////////////////////////////////////////////////////////////////////////// 293/////////////////////////////////////////////////////////////////////////////
108// encoder 294// encoder
109 295
110// structure used for encoding JSON 296// structure used for encoding JSON
111typedef struct 297typedef struct
… …
113 char *cur; // SvPVX (sv) + current output position 299 char *cur; // SvPVX (sv) + current output position
114 char *end; // SvEND (sv) 300 char *end; // SvEND (sv)
115 SV *sv; // result scalar 301 SV *sv; // result scalar
116 JSON json; 302 JSON json;
117 U32 indent; // indentation level 303 U32 indent; // indentation level
118 U32 maxdepth; // max. indentation/recursion level 304 UV limit; // escape character values >= this value when encoding
119} enc_t; 305} enc_t;
120 306
121inline void 307INLINE void
122need (enc_t *enc, STRLEN len) 308need (enc_t *enc, STRLEN len)
123{ 309{
124 if (expect_false (enc->cur + len >= enc->end)) 310 if (expect_false (enc->cur + len >= enc->end))
125 { 311 {
126 STRLEN cur = enc->cur - SvPVX (enc->sv); 312 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
127 SvGROW (enc->sv, cur + len + 1); 313 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
128 enc->cur = SvPVX (enc->sv) + cur; 314 enc->cur = SvPVX (enc->sv) + cur;
129 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 315 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
130 } 316 }
131} 317}
132 318
133inline void 319INLINE void
134encode_ch (enc_t *enc, char ch) 320encode_ch (enc_t *enc, char ch)
135{ 321{
136 need (enc, 1); 322 need (enc, 1);
137 *enc->cur++ = ch; 323 *enc->cur++ = ch;
138} 324}
… …
192 { 378 {
193 uch = ch; 379 uch = ch;
194 clen = 1; 380 clen = 1;
195 } 381 }
196 382
197 if (uch > 0x10FFFFUL) 383 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
198 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
199
200 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
201 { 384 {
202 if (uch > 0xFFFFUL) 385 if (uch >= 0x10000UL)
203 { 386 {
387 if (uch >= 0x110000UL)
388 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
389
204 need (enc, len += 11); 390 need (enc, len += 11);
205 sprintf (enc->cur, "\\u%04x\\u%04x", 391 sprintf (enc->cur, "\\u%04x\\u%04x",
206 (int)((uch - 0x10000) / 0x400 + 0xD800), 392 (int)((uch - 0x10000) / 0x400 + 0xD800),
207 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 393 (int)((uch - 0x10000) % 0x400 + 0xDC00));
208 enc->cur += 12; 394 enc->cur += 12;
209 } 395 }
210 else 396 else
211 { 397 {
212 static char hexdigit [16] = "0123456789abcdef";
213 need (enc, len += 5); 398 need (enc, len += 5);
214 *enc->cur++ = '\\'; 399 *enc->cur++ = '\\';
215 *enc->cur++ = 'u'; 400 *enc->cur++ = 'u';
216 *enc->cur++ = hexdigit [ uch >> 12 ]; 401 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
217 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 402 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
218 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 403 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
219 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 404 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
220 } 405 }
221 406
222 str += clen; 407 str += clen;
223 } 408 }
224 else if (enc->json.flags & F_LATIN1) 409 else if (enc->json.flags & F_LATIN1)
… …
236 while (--clen); 421 while (--clen);
237 } 422 }
238 else 423 else
239 { 424 {
240 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 425 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
241 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 426 enc->cur = encode_utf8 (enc->cur, uch);
242 ++str; 427 ++str;
243 } 428 }
244 } 429 }
245 } 430 }
246 } 431 }
247 432
248 --len; 433 --len;
249 } 434 }
250} 435}
251 436
252inline void 437INLINE void
253encode_indent (enc_t *enc) 438encode_indent (enc_t *enc)
254{ 439{
255 if (enc->json.flags & F_INDENT) 440 if (enc->json.flags & F_INDENT)
256 { 441 {
257 int spaces = enc->indent * INDENT_STEP; 442 int spaces = enc->indent * INDENT_STEP;
… …
260 memset (enc->cur, ' ', spaces); 445 memset (enc->cur, ' ', spaces);
261 enc->cur += spaces; 446 enc->cur += spaces;
262 } 447 }
263} 448}
264 449
265inline void 450INLINE void
266encode_space (enc_t *enc) 451encode_space (enc_t *enc)
267{ 452{
268 need (enc, 1); 453 need (enc, 1);
269 encode_ch (enc, ' '); 454 encode_ch (enc, ' ');
270} 455}
271 456
272inline void 457INLINE void
273encode_nl (enc_t *enc) 458encode_nl (enc_t *enc)
274{ 459{
275 if (enc->json.flags & F_INDENT) 460 if (enc->json.flags & F_INDENT)
276 { 461 {
277 need (enc, 1); 462 need (enc, 1);
278 encode_ch (enc, '\n'); 463 encode_ch (enc, '\n');
279 } 464 }
280} 465}
281 466
282inline void 467INLINE void
283encode_comma (enc_t *enc) 468encode_comma (enc_t *enc)
284{ 469{
285 encode_ch (enc, ','); 470 encode_ch (enc, ',');
286 471
287 if (enc->json.flags & F_INDENT) 472 if (enc->json.flags & F_INDENT)
… …
295static void 480static void
296encode_av (enc_t *enc, AV *av) 481encode_av (enc_t *enc, AV *av)
297{ 482{
298 int i, len = av_len (av); 483 int i, len = av_len (av);
299 484
300 if (enc->indent >= enc->maxdepth) 485 if (enc->indent >= enc->json.max_depth)
301 croak ("data structure too deep (hit recursion limit)"); 486 croak (ERR_NESTING_EXCEEDED);
302 487
303 encode_ch (enc, '['); encode_nl (enc); 488 encode_ch (enc, '[');
304 ++enc->indent;
305 489
490 if (len >= 0)
491 {
492 encode_nl (enc); ++enc->indent;
493
306 for (i = 0; i <= len; ++i) 494 for (i = 0; i <= len; ++i)
307 { 495 {
496 SV **svp = av_fetch (av, i, 0);
497
308 encode_indent (enc); 498 encode_indent (enc);
309 encode_sv (enc, *av_fetch (av, i, 0));
310 499
500 if (svp)
501 encode_sv (enc, *svp);
502 else
503 encode_str (enc, "null", 4, 0);
504
311 if (i < len) 505 if (i < len)
312 encode_comma (enc); 506 encode_comma (enc);
313 } 507 }
314 508
509 encode_nl (enc); --enc->indent; encode_indent (enc);
510 }
511
315 encode_nl (enc); 512 encode_ch (enc, ']');
316
317 --enc->indent;
318 encode_indent (enc); encode_ch (enc, ']');
319} 513}
320 514
321static void 515static void
322encode_he (enc_t *enc, HE *he) 516encode_hk (enc_t *enc, HE *he)
323{ 517{
324 encode_ch (enc, '"'); 518 encode_ch (enc, '"');
325 519
326 if (HeKLEN (he) == HEf_SVKEY) 520 if (HeKLEN (he) == HEf_SVKEY)
327 { 521 {
328 SV *sv = HeSVKEY (he); 522 SV *sv = HeSVKEY (he);
329 STRLEN len; 523 STRLEN len;
330 char *str; 524 char *str;
331 525
332 SvGETMAGIC (sv); 526 SvGETMAGIC (sv);
333 str = SvPV (sv, len); 527 str = SvPV (sv, len);
334 528
335 encode_str (enc, str, len, SvUTF8 (sv)); 529 encode_str (enc, str, len, SvUTF8 (sv));
336 } 530 }
… …
340 encode_ch (enc, '"'); 534 encode_ch (enc, '"');
341 535
342 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); 536 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
343 encode_ch (enc, ':'); 537 encode_ch (enc, ':');
344 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); 538 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
345 encode_sv (enc, HeVAL (he));
346} 539}
347 540
348// compare hash entries, used when all keys are bytestrings 541// compare hash entries, used when all keys are bytestrings
349static int 542static int
350he_cmp_fast (const void *a_, const void *b_) 543he_cmp_fast (const void *a_, const void *b_)
… …
355 HE *b = *(HE **)b_; 548 HE *b = *(HE **)b_;
356 549
357 STRLEN la = HeKLEN (a); 550 STRLEN la = HeKLEN (a);
358 STRLEN lb = HeKLEN (b); 551 STRLEN lb = HeKLEN (b);
359 552
360 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 553 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
361 cmp = la - lb; 554 cmp = lb - la;
362 555
363 return cmp; 556 return cmp;
364} 557}
365 558
366// compare hash entries, used when some keys are sv's or utf-x 559// compare hash entries, used when some keys are sv's or utf-x
367static int 560static int
368he_cmp_slow (const void *a, const void *b) 561he_cmp_slow (const void *a, const void *b)
369{ 562{
370 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 563 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
371} 564}
372 565
373static void 566static void
374encode_hv (enc_t *enc, HV *hv) 567encode_hv (enc_t *enc, HV *hv)
375{ 568{
376 int count, i; 569 HE *he;
377 570
378 if (enc->indent >= enc->maxdepth) 571 if (enc->indent >= enc->json.max_depth)
379 croak ("data structure too deep (hit recursion limit)"); 572 croak (ERR_NESTING_EXCEEDED);
380 573
381 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 574 encode_ch (enc, '{');
382 575
383 if ((count = hv_iterinit (hv)))
384 {
385 // for canonical output we have to sort by keys first 576 // for canonical output we have to sort by keys first
386 // actually, this is mostly due to the stupid so-called 577 // actually, this is mostly due to the stupid so-called
387 // security workaround added somewhere in 5.8.x. 578 // security workaround added somewhere in 5.8.x
388 // that randomises hash orderings 579 // that randomises hash orderings
389 if (enc->json.flags & F_CANONICAL) 580 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
581 {
582 int count = hv_iterinit (hv);
583
584 if (SvMAGICAL (hv))
390 { 585 {
586 // need to count by iterating. could improve by dynamically building the vector below
587 // but I don't care for the speed of this special case.
588 // note also that we will run into undefined behaviour when the two iterations
589 // do not result in the same count, something I might care for in some later release.
590
591 count = 0;
592 while (hv_iternext (hv))
593 ++count;
594
595 hv_iterinit (hv);
596 }
597
598 if (count)
599 {
391 int fast = 1; 600 int i, fast = 1;
392 HE *he; 601 HE *hes_stack [STACK_HES];
393#if defined(__BORLANDC__) || defined(_MSC_VER) 602 HE **hes = hes_stack;
394 HE **hes = _alloca (count * sizeof (HE)); 603
395#else 604 // allocate larger arrays on the heap
396 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 605 if (count > STACK_HES)
397#endif 606 {
607 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
608 hes = (HE **)SvPVX (sv);
609 }
398 610
399 i = 0; 611 i = 0;
400 while ((he = hv_iternext (hv))) 612 while ((he = hv_iternext (hv)))
401 { 613 {
402 hes [i++] = he; 614 hes [i++] = he;
… …
424 636
425 FREETMPS; 637 FREETMPS;
426 LEAVE; 638 LEAVE;
427 } 639 }
428 640
429 for (i = 0; i < count; ++i) 641 encode_nl (enc); ++enc->indent;
642
643 while (count--)
430 { 644 {
431 encode_indent (enc); 645 encode_indent (enc);
646 he = hes [count];
432 encode_he (enc, hes [i]); 647 encode_hk (enc, he);
648 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
433 649
434 if (i < count - 1) 650 if (count)
435 encode_comma (enc); 651 encode_comma (enc);
436 } 652 }
437 653
438 encode_nl (enc); 654 encode_nl (enc); --enc->indent; encode_indent (enc);
439 } 655 }
656 }
440 else 657 else
658 {
659 if (hv_iterinit (hv) || SvMAGICAL (hv))
660 if ((he = hv_iternext (hv)))
441 { 661 {
442 HE *he = hv_iternext (hv); 662 encode_nl (enc); ++enc->indent;
443 663
444 for (;;) 664 for (;;)
445 { 665 {
446 encode_indent (enc); 666 encode_indent (enc);
447 encode_he (enc, he); 667 encode_hk (enc, he);
668 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
448 669
449 if (!(he = hv_iternext (hv))) 670 if (!(he = hv_iternext (hv)))
450 break; 671 break;
451 672
452 encode_comma (enc); 673 encode_comma (enc);
453 } 674 }
454 675
455 encode_nl (enc); 676 encode_nl (enc); --enc->indent; encode_indent (enc);
456 } 677 }
457 } 678 }
458 679
459 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 680 encode_ch (enc, '}');
460} 681}
461 682
462// encode objects, arrays and special \0=false and \1=true values. 683// encode objects, arrays and special \0=false and \1=true values.
463static void 684static void
464encode_rv (enc_t *enc, SV *sv) 685encode_rv (enc_t *enc, SV *sv)
… …
468 SvGETMAGIC (sv); 689 SvGETMAGIC (sv);
469 svt = SvTYPE (sv); 690 svt = SvTYPE (sv);
470 691
471 if (expect_false (SvOBJECT (sv))) 692 if (expect_false (SvOBJECT (sv)))
472 { 693 {
473 if (SvSTASH (sv) == json_boolean_stash) 694 HV *boolean_stash = !JSON_SLOW || types_boolean_stash
695 ? types_boolean_stash
696 : gv_stashpv ("Types::Serialiser::Boolean", 1);
697 HV *stash = SvSTASH (sv);
698
699 if (stash == boolean_stash)
474 { 700 {
475 if (SvIV (sv) == 0) 701 if (SvIV (sv))
702 encode_str (enc, "true", 4, 0);
703 else
476 encode_str (enc, "false", 5, 0); 704 encode_str (enc, "false", 5, 0);
477 else
478 encode_str (enc, "true", 4, 0);
479 } 705 }
480 else 706 else if (enc->json.flags & F_CONV_BLESSED)
481 { 707 {
482#if 0 708 GV *to_json = gv_fetchmethod_autoload (stash, "TO_JSON", 0);
483 if (0 && sv_derived_from (rv, "JSON::Literal")) 709
484 { 710 if (to_json)
485 // not yet
486 } 711 {
487#endif
488 if (enc->json.flags & F_CONV_BLESSED)
489 { 712 dSP;
713
714 ENTER; SAVETMPS; PUSHMARK (SP);
490 // we re-bless the reference to get overload and other niceties right 715 // we re-bless the reference to get overload and other niceties right
491 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
492
493 if (to_json)
494 {
495 dSP;
496 ENTER;
497 SAVETMPS;
498 PUSHMARK (SP);
499 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 716 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
500 717
501 // calling with G_SCALAR ensures that we always get a 1 reutrn value 718 // calling with G_SCALAR ensures that we always get a 1 return value
502 // check anyways.
503 PUTBACK; 719 PUTBACK;
504 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR)); 720 call_sv ((SV *)GvCV (to_json), G_SCALAR);
505 SPAGAIN; 721 SPAGAIN;
506 722
723 // catch this surprisingly common error
724 if (SvROK (TOPs) && SvRV (TOPs) == sv)
725 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
726
727 sv = POPs;
728 PUTBACK;
729
507 encode_sv (enc, POPs); 730 encode_sv (enc, sv);
508 731
509 FREETMPS; 732 FREETMPS; LEAVE;
510 LEAVE;
511 }
512 else if (enc->json.flags & F_ALLOW_BLESSED)
513 encode_str (enc, "null", 4, 0);
514 else
515 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
516 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
517 } 733 }
518 else if (enc->json.flags & F_ALLOW_BLESSED) 734 else if (enc->json.flags & F_ALLOW_BLESSED)
519 encode_str (enc, "null", 4, 0); 735 encode_str (enc, "null", 4, 0);
520 else 736 else
521 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 737 croak ("encountered object '%s' when conv_object is enabled, but TO_JSON is missing and allow_blessed disabled",
522 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 738 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
523 } 739 }
740 else if (enc->json.flags & F_ALLOW_TAGS)
741 {
742 GV *method = gv_fetchmethod_autoload (stash, "FREEZE", 0);
743
744 if (method)
745 {
746 int count;
747 dSP;
748
749 ENTER; SAVETMPS; PUSHMARK (SP);
750 EXTEND (SP, 2);
751 // we re-bless the reference to get overload and other niceties right
752 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
753 PUSHs (sv_json);
754
755 PUTBACK;
756 count = call_sv ((SV *)GvCV (method), G_ARRAY);
757 SPAGAIN;
758
759 // catch this surprisingly common error
760 if (SvROK (TOPs) && SvRV (TOPs) == sv)
761 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
762
763 encode_ch (enc, '(');
764 encode_ch (enc, '"');
765 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
766 encode_ch (enc, '"');
767 encode_ch (enc, ')');
768 encode_ch (enc, '[');
769
770 while (count)
771 {
772 encode_sv (enc, SP[1 - count--]);
773
774 if (count)
775 encode_ch (enc, ',');
776 }
777
778 encode_ch (enc, ']');
779
780 PUTBACK;
781
782 FREETMPS; LEAVE;
783 }
784 else if (enc->json.flags & F_ALLOW_BLESSED)
785 encode_str (enc, "null", 4, 0);
786 else
787 croak ("encountered object '%s' when allow_tags is enabled, but FREEZE is missing and allow_blessed disabled",
788 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
789 }
790 else if (enc->json.flags & F_ALLOW_BLESSED)
791 encode_str (enc, "null", 4, 0);
792 else
793 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled",
794 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
524 } 795 }
525 else if (svt == SVt_PVHV) 796 else if (svt == SVt_PVHV)
526 encode_hv (enc, (HV *)sv); 797 encode_hv (enc, (HV *)sv);
527 else if (svt == SVt_PVAV) 798 else if (svt == SVt_PVAV)
528 encode_av (enc, (AV *)sv); 799 encode_av (enc, (AV *)sv);
529 else if (svt < SVt_PVAV) 800 else if (svt < SVt_PVAV)
530 { 801 {
531 if (SvNIOK (sv) && SvIV (sv) == 0) 802 STRLEN len = 0;
803 char *pv = svt ? SvPV (sv, len) : 0;
804
805 if (len == 1 && *pv == '1')
806 encode_str (enc, "true", 4, 0);
807 else if (len == 1 && *pv == '0')
532 encode_str (enc, "false", 5, 0); 808 encode_str (enc, "false", 5, 0);
533 else if (SvNIOK (sv) && SvIV (sv) == 1) 809 else if (enc->json.flags & F_ALLOW_UNKNOWN)
534 encode_str (enc, "true", 4, 0); 810 encode_str (enc, "null", 4, 0);
535 else 811 else
536 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 812 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
537 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 813 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
538 } 814 }
815 else if (enc->json.flags & F_ALLOW_UNKNOWN)
816 encode_str (enc, "null", 4, 0);
539 else 817 else
540 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 818 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
541 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 819 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
542} 820}
543 821
… …
561 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 839 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
562 enc->cur += strlen (enc->cur); 840 enc->cur += strlen (enc->cur);
563 } 841 }
564 else if (SvIOKp (sv)) 842 else if (SvIOKp (sv))
565 { 843 {
566 // we assume we can always read an IV as a UV 844 // we assume we can always read an IV as a UV and vice versa
567 if (SvUV (sv) & ~(UV)0x7fff) 845 // we assume two's complement
568 { 846 // we assume no aliasing issues in the union
569 // large integer, use the (rather slow) snprintf way. 847 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
570 need (enc, sizeof (UV) * 3); 848 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
571 enc->cur +=
572 SvIsUV(sv)
573 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
574 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
575 }
576 else
577 { 849 {
578 // optimise the "small number case" 850 // optimise the "small number case"
579 // code will likely be branchless and use only a single multiplication 851 // code will likely be branchless and use only a single multiplication
852 // works for numbers up to 59074
580 I32 i = SvIV (sv); 853 I32 i = SvIVX (sv);
581 U32 u; 854 U32 u;
582 char digit, nz = 0; 855 char digit, nz = 0;
583 856
584 need (enc, 6); 857 need (enc, 6);
585 858
… …
591 864
592 // now output digit by digit, each time masking out the integer part 865 // now output digit by digit, each time masking out the integer part
593 // and multiplying by 5 while moving the decimal point one to the right, 866 // and multiplying by 5 while moving the decimal point one to the right,
594 // resulting in a net multiplication by 10. 867 // resulting in a net multiplication by 10.
595 // we always write the digit to memory but conditionally increment 868 // we always write the digit to memory but conditionally increment
596 // the pointer, to ease the usage of conditional move instructions. 869 // the pointer, to enable the use of conditional move instructions.
597 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 870 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
598 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 871 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
599 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 872 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
600 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 873 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
601 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 874 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
602 } 875 }
876 else
877 {
878 // large integer, use the (rather slow) snprintf way.
879 need (enc, IVUV_MAXCHARS);
880 enc->cur +=
881 SvIsUV(sv)
882 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
883 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
884 }
603 } 885 }
604 else if (SvROK (sv)) 886 else if (SvROK (sv))
605 encode_rv (enc, SvRV (sv)); 887 encode_rv (enc, SvRV (sv));
606 else if (!SvOK (sv)) 888 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
607 encode_str (enc, "null", 4, 0); 889 encode_str (enc, "null", 4, 0);
608 else 890 else
609 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 891 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
610 SvPV_nolen (sv), SvFLAGS (sv)); 892 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
611} 893}
612 894
613static SV * 895static SV *
614encode_json (SV *scalar, JSON *json) 896encode_json (SV *scalar, JSON *json)
615{ 897{
… …
621 enc.json = *json; 903 enc.json = *json;
622 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 904 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
623 enc.cur = SvPVX (enc.sv); 905 enc.cur = SvPVX (enc.sv);
624 enc.end = SvEND (enc.sv); 906 enc.end = SvEND (enc.sv);
625 enc.indent = 0; 907 enc.indent = 0;
626 enc.maxdepth = DEC_DEPTH (enc.json.flags); 908 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
909 : enc.json.flags & F_LATIN1 ? 0x000100UL
910 : 0x110000UL;
627 911
628 SvPOK_only (enc.sv); 912 SvPOK_only (enc.sv);
629 encode_sv (&enc, scalar); 913 encode_sv (&enc, scalar);
914 encode_nl (&enc);
630 915
631 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 916 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
632 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 917 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
633 918
634 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 919 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
652 JSON json; 937 JSON json;
653 U32 depth; // recursion depth 938 U32 depth; // recursion depth
654 U32 maxdepth; // recursion depth limit 939 U32 maxdepth; // recursion depth limit
655} dec_t; 940} dec_t;
656 941
657inline void 942INLINE void
943decode_comment (dec_t *dec)
944{
945 // only '#'-style comments allowed a.t.m.
946
947 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
948 ++dec->cur;
949}
950
951INLINE void
658decode_ws (dec_t *dec) 952decode_ws (dec_t *dec)
659{ 953{
660 for (;;) 954 for (;;)
661 { 955 {
662 char ch = *dec->cur; 956 char ch = *dec->cur;
663 957
664 if (ch > 0x20 958 if (ch > 0x20)
959 {
960 if (expect_false (ch == '#'))
961 {
962 if (dec->json.flags & F_RELAXED)
963 decode_comment (dec);
964 else
965 break;
966 }
967 else
968 break;
969 }
665 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 970 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
666 break; 971 break; // parse error, but let higher level handle it, gives better error messages
667 972
668 ++dec->cur; 973 ++dec->cur;
669 } 974 }
670} 975}
671 976
… …
675 if (*dec->cur != ch) \ 980 if (*dec->cur != ch) \
676 ERR (# ch " expected"); \ 981 ERR (# ch " expected"); \
677 ++dec->cur; \ 982 ++dec->cur; \
678 SE 983 SE
679 984
680#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 985#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
681#define DEC_DEC_DEPTH --dec->depth 986#define DEC_DEC_DEPTH --dec->depth
682 987
683static SV *decode_sv (dec_t *dec); 988static SV *decode_sv (dec_t *dec);
684 989
685static signed char decode_hexdigit[256]; 990static signed char decode_hexdigit[256];
… …
777 1082
778 if (hi >= 0x80) 1083 if (hi >= 0x80)
779 { 1084 {
780 utf8 = 1; 1085 utf8 = 1;
781 1086
782 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1087 cur = encode_utf8 (cur, hi);
783 } 1088 }
784 else 1089 else
785 *cur++ = hi; 1090 *cur++ = hi;
786 } 1091 }
787 break; 1092 break;
… …
789 default: 1094 default:
790 --dec_cur; 1095 --dec_cur;
791 ERR ("illegal backslash escape sequence in string"); 1096 ERR ("illegal backslash escape sequence in string");
792 } 1097 }
793 } 1098 }
794 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1099 else if (expect_true (ch >= 0x20 && ch < 0x80))
795 *cur++ = ch; 1100 *cur++ = ch;
796 else if (ch >= 0x80) 1101 else if (ch >= 0x80)
797 { 1102 {
798 STRLEN clen; 1103 STRLEN clen;
799 UV uch;
800 1104
801 --dec_cur; 1105 --dec_cur;
802 1106
803 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1107 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
804 if (clen == (STRLEN)-1) 1108 if (clen == (STRLEN)-1)
805 ERR ("malformed UTF-8 character in JSON string"); 1109 ERR ("malformed UTF-8 character in JSON string");
806 1110
807 do 1111 do
808 *cur++ = *dec_cur++; 1112 *cur++ = *dec_cur++;
… …
825 { 1129 {
826 STRLEN len = cur - buf; 1130 STRLEN len = cur - buf;
827 1131
828 if (sv) 1132 if (sv)
829 { 1133 {
830 SvGROW (sv, SvCUR (sv) + len + 1); 1134 STRLEN cur = SvCUR (sv);
1135
1136 if (SvLEN (sv) <= cur + len)
1137 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1138
831 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1139 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
832 SvCUR_set (sv, SvCUR (sv) + len); 1140 SvCUR_set (sv, SvCUR (sv) + len);
833 } 1141 }
834 else 1142 else
835 sv = newSVpvn (buf, len); 1143 sv = newSVpvn (buf, len);
… …
920 is_nv = 1; 1228 is_nv = 1;
921 } 1229 }
922 1230
923 if (!is_nv) 1231 if (!is_nv)
924 { 1232 {
1233 int len = dec->cur - start;
1234
925 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1235 // special case the rather common 1..5-digit-int case
926 if (*start == '-') 1236 if (*start == '-')
927 switch (dec->cur - start) 1237 switch (len)
928 { 1238 {
929 case 2: return newSViv (-( start [1] - '0' * 1)); 1239 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
930 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1240 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
931 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1241 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
932 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1242 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1243 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
933 } 1244 }
934 else 1245 else
935 switch (dec->cur - start) 1246 switch (len)
936 { 1247 {
937 case 1: return newSViv ( start [0] - '0' * 1); 1248 case 1: return newSViv ( start [0] - '0' * 1);
938 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1249 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
939 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1250 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
940 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1251 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1252 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
941 } 1253 }
942 1254
943 { 1255 {
944 UV uv; 1256 UV uv;
945 int numtype = grok_number (start, dec->cur - start, &uv); 1257 int numtype = grok_number (start, len, &uv);
946 if (numtype & IS_NUMBER_IN_UV) 1258 if (numtype & IS_NUMBER_IN_UV)
947 if (numtype & IS_NUMBER_NEG) 1259 if (numtype & IS_NUMBER_NEG)
948 { 1260 {
949 if (uv < (UV)IV_MIN) 1261 if (uv < (UV)IV_MIN)
950 return newSViv (-(IV)uv); 1262 return newSViv (-(IV)uv);
951 } 1263 }
952 else 1264 else
953 return newSVuv (uv); 1265 return newSVuv (uv);
954
955 // here would likely be the place for bigint support
956 } 1266 }
957 }
958 1267
959 // if we ever support bigint or bigfloat, this is the place for bigfloat 1268 len -= *start == '-' ? 1 : 0;
1269
1270 // does not fit into IV or UV, try NV
1271 if (len <= NV_DIG)
1272 // fits into NV without loss of precision
1273 return newSVnv (json_atof (start));
1274
1275 // everything else fails, convert it to a string
1276 return newSVpvn (start, dec->cur - start);
1277 }
1278
1279 // loss of precision here
960 return newSVnv (Atof (start)); 1280 return newSVnv (json_atof (start));
961 1281
962fail: 1282fail:
963 return 0; 1283 return 0;
964} 1284}
965 1285
… …
989 if (*dec->cur == ']') 1309 if (*dec->cur == ']')
990 { 1310 {
991 ++dec->cur; 1311 ++dec->cur;
992 break; 1312 break;
993 } 1313 }
994 1314
995 if (*dec->cur != ',') 1315 if (*dec->cur != ',')
996 ERR (", or ] expected while parsing array"); 1316 ERR (", or ] expected while parsing array");
997 1317
998 ++dec->cur; 1318 ++dec->cur;
1319
1320 decode_ws (dec);
1321
1322 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1323 {
1324 ++dec->cur;
1325 break;
1326 }
999 } 1327 }
1000 1328
1001 DEC_DEC_DEPTH; 1329 DEC_DEC_DEPTH;
1002 return newRV_noinc ((SV *)av); 1330 return newRV_noinc ((SV *)av);
1003 1331
… …
1008} 1336}
1009 1337
1010static SV * 1338static SV *
1011decode_hv (dec_t *dec) 1339decode_hv (dec_t *dec)
1012{ 1340{
1341 SV *sv;
1013 HV *hv = newHV (); 1342 HV *hv = newHV ();
1014 1343
1015 DEC_INC_DEPTH; 1344 DEC_INC_DEPTH;
1016 decode_ws (dec); 1345 decode_ws (dec);
1017 1346
1018 if (*dec->cur == '}') 1347 if (*dec->cur == '}')
1019 ++dec->cur; 1348 ++dec->cur;
1020 else 1349 else
1021 for (;;) 1350 for (;;)
1022 { 1351 {
1023 decode_ws (dec); EXPECT_CH ('"'); 1352 EXPECT_CH ('"');
1024 1353
1025 // heuristic: assume that 1354 // heuristic: assume that
1026 // a) decode_str + hv_store_ent are abysmally slow. 1355 // a) decode_str + hv_store_ent are abysmally slow.
1027 // b) most hash keys are short, simple ascii text. 1356 // b) most hash keys are short, simple ascii text.
1028 // => try to "fast-match" such strings to avoid 1357 // => try to "fast-match" such strings to avoid
… …
1032 char *p = dec->cur; 1361 char *p = dec->cur;
1033 char *e = p + 24; // only try up to 24 bytes 1362 char *e = p + 24; // only try up to 24 bytes
1034 1363
1035 for (;;) 1364 for (;;)
1036 { 1365 {
1037 // the >= 0x80 is true on most architectures 1366 // the >= 0x80 is false on most architectures
1038 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1367 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1039 { 1368 {
1040 // slow path, back up and use decode_str 1369 // slow path, back up and use decode_str
1041 SV *key = decode_str (dec); 1370 SV *key = decode_str (dec);
1042 if (!key) 1371 if (!key)
1043 goto fail; 1372 goto fail;
1044 1373
1045 decode_ws (dec); EXPECT_CH (':'); 1374 decode_ws (dec); EXPECT_CH (':');
1046 1375
1376 decode_ws (dec);
1047 value = decode_sv (dec); 1377 value = decode_sv (dec);
1048 if (!value) 1378 if (!value)
1049 { 1379 {
1050 SvREFCNT_dec (key); 1380 SvREFCNT_dec (key);
1051 goto fail; 1381 goto fail;
… …
1063 int len = p - key; 1393 int len = p - key;
1064 dec->cur = p + 1; 1394 dec->cur = p + 1;
1065 1395
1066 decode_ws (dec); EXPECT_CH (':'); 1396 decode_ws (dec); EXPECT_CH (':');
1067 1397
1398 decode_ws (dec);
1068 value = decode_sv (dec); 1399 value = decode_sv (dec);
1069 if (!value) 1400 if (!value)
1070 goto fail; 1401 goto fail;
1071 1402
1072 hv_store (hv, key, len, value, 0); 1403 hv_store (hv, key, len, value, 0);
… …
1088 1419
1089 if (*dec->cur != ',') 1420 if (*dec->cur != ',')
1090 ERR (", or } expected while parsing object/hash"); 1421 ERR (", or } expected while parsing object/hash");
1091 1422
1092 ++dec->cur; 1423 ++dec->cur;
1424
1425 decode_ws (dec);
1426
1427 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1428 {
1429 ++dec->cur;
1430 break;
1431 }
1093 } 1432 }
1094 1433
1095 DEC_DEC_DEPTH; 1434 DEC_DEC_DEPTH;
1096 return newRV_noinc ((SV *)hv); 1435 sv = newRV_noinc ((SV *)hv);
1436
1437 // check filter callbacks
1438 if (dec->json.flags & F_HOOK)
1439 {
1440 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1441 {
1442 HE *cb, *he;
1443
1444 hv_iterinit (hv);
1445 he = hv_iternext (hv);
1446 hv_iterinit (hv);
1447
1448 // the next line creates a mortal sv each time its called.
1449 // might want to optimise this for common cases.
1450 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1451
1452 if (cb)
1453 {
1454 dSP;
1455 int count;
1456
1457 ENTER; SAVETMPS; PUSHMARK (SP);
1458 XPUSHs (HeVAL (he));
1459 sv_2mortal (sv);
1460
1461 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1462
1463 if (count == 1)
1464 {
1465 sv = newSVsv (POPs);
1466 FREETMPS; LEAVE;
1467 return sv;
1468 }
1469
1470 SvREFCNT_inc (sv);
1471 FREETMPS; LEAVE;
1472 }
1473 }
1474
1475 if (dec->json.cb_object)
1476 {
1477 dSP;
1478 int count;
1479
1480 ENTER; SAVETMPS; PUSHMARK (SP);
1481 XPUSHs (sv_2mortal (sv));
1482
1483 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1484
1485 if (count == 1)
1486 {
1487 sv = newSVsv (POPs);
1488 FREETMPS; LEAVE;
1489 return sv;
1490 }
1491
1492 SvREFCNT_inc (sv);
1493 FREETMPS; LEAVE;
1494 }
1495 }
1496
1497 return sv;
1097 1498
1098fail: 1499fail:
1099 SvREFCNT_dec (hv); 1500 SvREFCNT_dec (hv);
1100 DEC_DEC_DEPTH; 1501 DEC_DEC_DEPTH;
1101 return 0; 1502 return 0;
1102} 1503}
1103 1504
1104static SV * 1505static SV *
1506decode_tag (dec_t *dec)
1507{
1508 SV *tag = 0;
1509 SV *val = 0;
1510
1511 if (!(dec->json.flags & F_ALLOW_TAGS))
1512 ERR ("malformed JSON string, neither array, object, number, string or atom");
1513
1514 ++dec->cur;
1515
1516 tag = decode_sv (dec);
1517 if (!tag)
1518 goto fail;
1519
1520 if (!SvPOK (tag))
1521 ERR ("malformed JSON string, (tag) must be a string");
1522
1523 if (*dec->cur != ')')
1524 ERR (") expected after tag");
1525
1526 ++dec->cur;
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 *
1105decode_sv (dec_t *dec) 1582decode_sv (dec_t *dec)
1106{ 1583{
1107 decode_ws (dec);
1108
1109 // the beauty of JSON: you need exactly one character lookahead 1584 // the beauty of JSON: you need exactly one character lookahead
1110 // to parse anything. 1585 // to parse everything.
1111 switch (*dec->cur) 1586 switch (*dec->cur)
1112 { 1587 {
1113 case '"': ++dec->cur; return decode_str (dec); 1588 case '"': ++dec->cur; return decode_str (dec);
1114 case '[': ++dec->cur; return decode_av (dec); 1589 case '[': ++dec->cur; return decode_av (dec);
1115 case '{': ++dec->cur; return decode_hv (dec); 1590 case '{': ++dec->cur; return decode_hv (dec);
1591 case '(': return decode_tag (dec);
1116 1592
1117 case '-': 1593 case '-':
1118 case '0': case '1': case '2': case '3': case '4': 1594 case '0': case '1': case '2': case '3': case '4':
1119 case '5': case '6': case '7': case '8': case '9': 1595 case '5': case '6': case '7': case '8': case '9':
1120 return decode_num (dec); 1596 return decode_num (dec);
1121 1597
1122 case 't': 1598 case 't':
1123 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1599 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1124 { 1600 {
1125 dec->cur += 4; 1601 dec->cur += 4;
1126 return SvREFCNT_inc (json_true); 1602#if JSON_SLOW
1603 types_true = get_bool ("Types::Serialiser::true");
1604#endif
1605 return newSVsv (types_true);
1127 } 1606 }
1128 else 1607 else
1129 ERR ("'true' expected"); 1608 ERR ("'true' expected");
1130 1609
1131 break; 1610 break;
1132 1611
1133 case 'f': 1612 case 'f':
1134 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1613 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1135 { 1614 {
1136 dec->cur += 5; 1615 dec->cur += 5;
1137 return SvREFCNT_inc (json_false); 1616#if JSON_SLOW
1617 types_false = get_bool ("Types::Serialiser::false");
1618#endif
1619 return newSVsv (types_false);
1138 } 1620 }
1139 else 1621 else
1140 ERR ("'false' expected"); 1622 ERR ("'false' expected");
1141 1623
1142 break; 1624 break;
… …
1151 ERR ("'null' expected"); 1633 ERR ("'null' expected");
1152 1634
1153 break; 1635 break;
1154 1636
1155 default: 1637 default:
1156 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1638 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1157 break; 1639 break;
1158 } 1640 }
1159 1641
1160fail: 1642fail:
1161 return 0; 1643 return 0;
1162} 1644}
1163 1645
1164static SV * 1646static SV *
1165decode_json (SV *string, JSON *json, UV *offset_return) 1647decode_json (SV *string, JSON *json, char **offset_return)
1166{ 1648{
1167 dec_t dec; 1649 dec_t dec;
1168 UV offset;
1169 SV *sv; 1650 SV *sv;
1170 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 */
1171 SvGETMAGIC (string); 1657 /*SvGETMAGIC (string);*/
1658 if (SvMAGICAL (string) || !SvPOK (string))
1659 string = sv_2mortal (newSVsv (string));
1660
1172 SvUPGRADE (string, SVt_PV); 1661 SvUPGRADE (string, SVt_PV);
1173 1662
1174 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)
1175 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",
1176 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1682 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1683 }
1177 1684
1178 if (json->flags & F_UTF8) 1685 if (DECODE_WANTS_OCTETS (json))
1179 sv_utf8_downgrade (string, 0); 1686 sv_utf8_downgrade (string, 0);
1180 else 1687 else
1181 sv_utf8_upgrade (string); 1688 sv_utf8_upgrade (string);
1182 1689
1183 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1690 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1184 1691
1185 dec.json = *json; 1692 dec.json = *json;
1186 dec.cur = SvPVX (string); 1693 dec.cur = SvPVX (string);
1187 dec.end = SvEND (string); 1694 dec.end = SvEND (string);
1188 dec.err = 0; 1695 dec.err = 0;
1189 dec.depth = 0; 1696 dec.depth = 0;
1190 dec.maxdepth = DEC_DEPTH (dec.json.flags); 1697
1698 if (dec.json.cb_object || dec.json.cb_sk_object)
1699 dec.json.flags |= F_HOOK;
1191 1700
1192 *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
1702
1703 decode_ws (&dec);
1193 sv = decode_sv (&dec); 1704 sv = decode_sv (&dec);
1705
1706 if (offset_return)
1707 *offset_return = dec.cur;
1194 1708
1195 if (!(offset_return || !sv)) 1709 if (!(offset_return || !sv))
1196 { 1710 {
1197 // check for trailing garbage 1711 // check for trailing garbage
1198 decode_ws (&dec); 1712 decode_ws (&dec);
… …
1201 { 1715 {
1202 dec.err = "garbage after JSON object"; 1716 dec.err = "garbage after JSON object";
1203 SvREFCNT_dec (sv); 1717 SvREFCNT_dec (sv);
1204 sv = 0; 1718 sv = 0;
1205 } 1719 }
1206 }
1207
1208 if (offset_return || !sv)
1209 {
1210 offset = dec.json.flags & F_UTF8
1211 ? dec.cur - SvPVX (string)
1212 : utf8_distance (dec.cur, SvPVX (string));
1213
1214 if (offset_return)
1215 *offset_return = offset;
1216 } 1720 }
1217 1721
1218 if (!sv) 1722 if (!sv)
1219 { 1723 {
1220 SV *uni = sv_newmortal (); 1724 SV *uni = sv_newmortal ();
… …
1226 SAVEVPTR (PL_curcop); 1730 SAVEVPTR (PL_curcop);
1227 PL_curcop = &cop; 1731 PL_curcop = &cop;
1228 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);
1229 LEAVE; 1733 LEAVE;
1230 1734
1231 croak ("%s, at character offset %d [\"%s\"]", 1735 croak ("%s, at character offset %d (before \"%s\")",
1232 dec.err, 1736 dec.err,
1233 (int)offset, 1737 (int)ptr_to_index (string, dec.cur),
1234 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1738 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1235 } 1739 }
1236 1740
1237 sv = sv_2mortal (sv); 1741 sv = sv_2mortal (sv);
1238 1742
1239 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1743 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1240 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)");
1241 1745
1242 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
1243} 1905}
1244 1906
1245///////////////////////////////////////////////////////////////////////////// 1907/////////////////////////////////////////////////////////////////////////////
1246// XS interface functions 1908// XS interface functions
1247 1909
… …
1256 i >= '0' && i <= '9' ? i - '0' 1918 i >= '0' && i <= '9' ? i - '0'
1257 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1919 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1258 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1920 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1259 : -1; 1921 : -1;
1260 1922
1261 json_stash = gv_stashpv ("JSON::XS" , 1); 1923 json_stash = gv_stashpv ("JSON::XS" , 1);
1262 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1924 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1263 1925
1264 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1926 types_true = get_bool ("Types::Serialiser::true");
1265 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 */
1266} 1933}
1267 1934
1268PROTOTYPES: DISABLE 1935PROTOTYPES: DISABLE
1936
1937void CLONE (...)
1938 CODE:
1939 json_stash = 0;
1940 types_boolean_stash = 0;
1269 1941
1270void new (char *klass) 1942void new (char *klass)
1271 PPCODE: 1943 PPCODE:
1272{ 1944{
1273 SV *pv = NEWSV (0, sizeof (JSON)); 1945 SV *pv = NEWSV (0, sizeof (JSON));
1274 SvPOK_only (pv); 1946 SvPOK_only (pv);
1275 Zero (SvPVX (pv), 1, sizeof (JSON)); 1947 json_init ((JSON *)SvPVX (pv));
1276 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1948 XPUSHs (sv_2mortal (sv_bless (
1277 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 )));
1278} 1952}
1279 1953
1280void ascii (JSON *self, int enable = 1) 1954void ascii (JSON *self, int enable = 1)
1281 ALIAS: 1955 ALIAS:
1282 ascii = F_ASCII 1956 ascii = F_ASCII
… …
1289 pretty = F_PRETTY 1963 pretty = F_PRETTY
1290 allow_nonref = F_ALLOW_NONREF 1964 allow_nonref = F_ALLOW_NONREF
1291 shrink = F_SHRINK 1965 shrink = F_SHRINK
1292 allow_blessed = F_ALLOW_BLESSED 1966 allow_blessed = F_ALLOW_BLESSED
1293 convert_blessed = F_CONV_BLESSED 1967 convert_blessed = F_CONV_BLESSED
1968 relaxed = F_RELAXED
1969 allow_unknown = F_ALLOW_UNKNOWN
1970 allow_tags = F_ALLOW_TAGS
1294 PPCODE: 1971 PPCODE:
1295{ 1972{
1296 if (enable) 1973 if (enable)
1297 self->flags |= ix; 1974 self->flags |= ix;
1298 else 1975 else
1299 self->flags &= ~ix; 1976 self->flags &= ~ix;
1300 1977
1301 XPUSHs (ST (0)); 1978 XPUSHs (ST (0));
1302} 1979}
1303 1980
1304void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1981void get_ascii (JSON *self)
1982 ALIAS:
1983 get_ascii = F_ASCII
1984 get_latin1 = F_LATIN1
1985 get_utf8 = F_UTF8
1986 get_indent = F_INDENT
1987 get_canonical = F_CANONICAL
1988 get_space_before = F_SPACE_BEFORE
1989 get_space_after = F_SPACE_AFTER
1990 get_allow_nonref = F_ALLOW_NONREF
1991 get_shrink = F_SHRINK
1992 get_allow_blessed = F_ALLOW_BLESSED
1993 get_convert_blessed = F_CONV_BLESSED
1994 get_relaxed = F_RELAXED
1995 get_allow_unknown = F_ALLOW_UNKNOWN
1996 get_allow_tags = F_ALLOW_TAGS
1305 PPCODE: 1997 PPCODE:
1306{ 1998 XPUSHs (boolSV (self->flags & ix));
1307 UV log2 = 0;
1308 1999
1309 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 2000void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1310 2001 PPCODE:
1311 while ((1UL << log2) < max_depth) 2002 self->max_depth = max_depth;
1312 ++log2;
1313
1314 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1315
1316 XPUSHs (ST (0)); 2003 XPUSHs (ST (0));
1317}
1318 2004
2005U32 get_max_depth (JSON *self)
2006 CODE:
2007 RETVAL = self->max_depth;
2008 OUTPUT:
2009 RETVAL
2010
1319void max_size (JSON *self, UV max_size = 0) 2011void max_size (JSON *self, U32 max_size = 0)
1320 PPCODE: 2012 PPCODE:
1321{ 2013 self->max_size = max_size;
1322 UV log2 = 0; 2014 XPUSHs (ST (0));
1323 2015
1324 if (max_size > 0x80000000UL) max_size = 0x80000000UL; 2016int get_max_size (JSON *self)
1325 if (max_size == 1) max_size = 2; 2017 CODE:
2018 RETVAL = self->max_size;
2019 OUTPUT:
2020 RETVAL
1326 2021
1327 while ((1UL << log2) < max_size) 2022void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1328 ++log2; 2023 PPCODE:
2024{
2025 SvREFCNT_dec (self->cb_object);
2026 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1329 2027
1330 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); 2028 XPUSHs (ST (0));
2029}
2030
2031void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
2032 PPCODE:
2033{
2034 if (!self->cb_sk_object)
2035 self->cb_sk_object = newHV ();
2036
2037 if (SvOK (cb))
2038 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
2039 else
2040 {
2041 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
2042
2043 if (!HvKEYS (self->cb_sk_object))
2044 {
2045 SvREFCNT_dec (self->cb_sk_object);
2046 self->cb_sk_object = 0;
2047 }
2048 }
1331 2049
1332 XPUSHs (ST (0)); 2050 XPUSHs (ST (0));
1333} 2051}
1334 2052
1335void encode (JSON *self, SV *scalar) 2053void encode (JSON *self, SV *scalar)
1336 PPCODE: 2054 PPCODE:
1337 XPUSHs (encode_json (scalar, self)); 2055 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2056 XPUSHs (scalar);
1338 2057
1339void decode (JSON *self, SV *jsonstr) 2058void decode (JSON *self, SV *jsonstr)
1340 PPCODE: 2059 PPCODE:
1341 XPUSHs (decode_json (jsonstr, self, 0)); 2060 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2061 XPUSHs (jsonstr);
1342 2062
1343void decode_prefix (JSON *self, SV *jsonstr) 2063void decode_prefix (JSON *self, SV *jsonstr)
1344 PPCODE: 2064 PPCODE:
1345{ 2065{
1346 UV offset; 2066 SV *sv;
2067 char *offset;
2068 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1347 EXTEND (SP, 2); 2069 EXTEND (SP, 2);
1348 PUSHs (decode_json (jsonstr, self, &offset)); 2070 PUSHs (sv);
1349 PUSHs (sv_2mortal (newSVuv (offset))); 2071 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1350} 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;
2195}
2196
2197void DESTROY (JSON *self)
2198 CODE:
2199 SvREFCNT_dec (self->cb_sk_object);
2200 SvREFCNT_dec (self->cb_object);
2201 SvREFCNT_dec (self->incr_text);
1351 2202
1352PROTOTYPES: ENABLE 2203PROTOTYPES: ENABLE
1353 2204
1354void to_json (SV *scalar) 2205void encode_json (SV *scalar)
1355 PPCODE: 2206 PPCODE:
1356{ 2207{
1357 JSON json = { F_DEFAULT | F_UTF8 }; 2208 JSON json;
1358 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);
1359} 2213}
1360 2214
1361void from_json (SV *jsonstr) 2215void decode_json (SV *jsonstr)
1362 PPCODE: 2216 PPCODE:
1363{ 2217{
1364 JSON json = { F_DEFAULT | F_UTF8 }; 2218 JSON json;
1365 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);
1366} 2223}
1367 2224

Diff Legend

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