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 1.35 by root, Wed Jun 6 14:52:49 2007 UTC vs.
Revision 1.121 by root, Tue Oct 29 00:06:40 2013 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include "stdio.h" 8#include <stdio.h>
9#include <limits.h>
10#include <float.h>
9 11
10#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 14#endif
13 15
16// some old perls do not have this, try to make it work, no
17// guarantees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13
20#endif
21
22// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
24
14#define F_ASCII 0x00000001UL 25#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL 26#define F_LATIN1 0x00000002UL
16#define F_UTF8 0x00000004UL 27#define F_UTF8 0x00000004UL
17#define F_INDENT 0x00000008UL 28#define F_INDENT 0x00000008UL
18#define F_CANONICAL 0x00000010UL 29#define F_CANONICAL 0x00000010UL
19#define F_SPACE_BEFORE 0x00000020UL 30#define F_SPACE_BEFORE 0x00000020UL
20#define F_SPACE_AFTER 0x00000040UL 31#define F_SPACE_AFTER 0x00000040UL
21#define F_ALLOW_NONREF 0x00000100UL 32#define F_ALLOW_NONREF 0x00000100UL
22#define F_SHRINK 0x00000200UL 33#define F_SHRINK 0x00000200UL
34#define F_ALLOW_BLESSED 0x00000400UL
35#define F_CONV_BLESSED 0x00000800UL
23#define F_MAXDEPTH 0xf8000000UL 36#define F_RELAXED 0x00001000UL
24#define S_MAXDEPTH 27 37#define F_ALLOW_UNKNOWN 0x00002000UL
25 38#define F_ALLOW_TAGS 0x00004000UL
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 39#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
27
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
30 40
31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
32#define F_DEFAULT (9UL << S_MAXDEPTH)
33 42
34#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
35#define INDENT_STEP 3 // spaces per indentation level 44#define INDENT_STEP 3 // spaces per indentation level
36 45
37#define SHORT_STRING_LEN 512 // special-case strings of up to this size 46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
47
48#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
38 49
39#define SB do { 50#define SB do {
40#define SE } while (0) 51#define SE } while (0)
41 52
42#if __GNUC__ >= 3 53#if __GNUC__ >= 3
43# define expect(expr,value) __builtin_expect ((expr),(value)) 54# define expect(expr,value) __builtin_expect ((expr), (value))
44# define inline inline 55# define INLINE static inline
45#else 56#else
46# define expect(expr,value) (expr) 57# define expect(expr,value) (expr)
47# define inline static 58# define INLINE static
48#endif 59#endif
49 60
50#define expect_false(expr) expect ((expr) != 0, 0) 61#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1) 62#define expect_true(expr) expect ((expr) != 0, 1)
52 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
53static HV *json_stash; // JSON::XS:: 81static HV *json_stash, *types_boolean_stash; // JSON::XS::
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)
94
95typedef struct {
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;
108} JSON;
109
110INLINE void
111json_init (JSON *json)
112{
113 Zero (json, 1, JSON);
114 json->max_depth = 512;
115}
54 116
55///////////////////////////////////////////////////////////////////////////// 117/////////////////////////////////////////////////////////////////////////////
56// utility functions 118// utility functions
57 119
58static UV * 120INLINE SV *
59SvJSON (SV *sv) 121get_bool (const char *name)
60{ 122{
61 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 123 SV *sv = get_sv (name, 1);
62 croak ("object is not of type JSON::XS");
63 124
64 return &SvUVX (SvRV (sv)); 125 SvREADONLY_on (sv);
65} 126 SvREADONLY_on (SvRV (sv));
66 127
67static void 128 return sv;
129}
130
131INLINE void
68shrink (SV *sv) 132shrink (SV *sv)
69{ 133{
70 sv_utf8_downgrade (sv, 1); 134 sv_utf8_downgrade (sv, 1);
135
71 if (SvLEN (sv) > SvCUR (sv) + 1) 136 if (SvLEN (sv) > SvCUR (sv) + 1)
72 { 137 {
73#ifdef SvPV_shrink_to_cur 138#ifdef SvPV_shrink_to_cur
74 SvPV_shrink_to_cur (sv); 139 SvPV_shrink_to_cur (sv);
75#elif defined (SvPV_renew) 140#elif defined (SvPV_renew)
81// 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
82// case of an error. 147// case of an error.
83// we special-case "safe" characters from U+80 .. U+7FF, 148// we special-case "safe" characters from U+80 .. U+7FF,
84// but use the very good perl function to parse anything else. 149// but use the very good perl function to parse anything else.
85// note that we never call this function for a ascii codepoints 150// note that we never call this function for a ascii codepoints
86inline UV 151INLINE UV
87decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 152decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
88{ 153{
89 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 154 if (expect_true (len >= 2
90 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 155 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
91 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 156 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
92 { 157 {
93 *clen = 2; 158 *clen = 2;
94 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 159 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
95 } 160 }
96 else 161 else
97 { 162 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
98 *clen = (STRLEN)-1; 163}
99 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 (;;)
100 } 215 {
101} 216 U8 dig = (U8)*s - '0';
102 217
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}
103///////////////////////////////////////////////////////////////////////////// 293/////////////////////////////////////////////////////////////////////////////
104// encoder 294// encoder
105 295
106// structure used for encoding JSON 296// structure used for encoding JSON
107typedef struct 297typedef struct
108{ 298{
109 char *cur; // SvPVX (sv) + current output position 299 char *cur; // SvPVX (sv) + current output position
110 char *end; // SvEND (sv) 300 char *end; // SvEND (sv)
111 SV *sv; // result scalar 301 SV *sv; // result scalar
112 U32 flags; // F_* 302 JSON json;
113 U32 indent; // indentation level 303 U32 indent; // indentation level
114 U32 maxdepth; // max. indentation/recursion level 304 UV limit; // escape character values >= this value when encoding
115} enc_t; 305} enc_t;
116 306
117inline void 307INLINE void
118need (enc_t *enc, STRLEN len) 308need (enc_t *enc, STRLEN len)
119{ 309{
120 if (expect_false (enc->cur + len >= enc->end)) 310 if (expect_false (enc->cur + len >= enc->end))
121 { 311 {
122 STRLEN cur = enc->cur - SvPVX (enc->sv); 312 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
123 SvGROW (enc->sv, cur + len + 1); 313 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
124 enc->cur = SvPVX (enc->sv) + cur; 314 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 315 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 316 }
127} 317}
128 318
129inline void 319INLINE void
130encode_ch (enc_t *enc, char ch) 320encode_ch (enc_t *enc, char ch)
131{ 321{
132 need (enc, 1); 322 need (enc, 1);
133 *enc->cur++ = ch; 323 *enc->cur++ = ch;
134} 324}
178 STRLEN clen; 368 STRLEN clen;
179 UV uch; 369 UV uch;
180 370
181 if (is_utf8) 371 if (is_utf8)
182 { 372 {
183 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184 uch = decode_utf8 (str, end - str, &clen); 373 uch = decode_utf8 (str, end - str, &clen);
185 if (clen == (STRLEN)-1) 374 if (clen == (STRLEN)-1)
186 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 375 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187 } 376 }
188 else 377 else
189 { 378 {
190 uch = ch; 379 uch = ch;
191 clen = 1; 380 clen = 1;
192 } 381 }
193 382
194 if (uch > 0x10FFFFUL) 383 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
195 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
196
197 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
198 { 384 {
199 if (uch > 0xFFFFUL) 385 if (uch >= 0x10000UL)
200 { 386 {
387 if (uch >= 0x110000UL)
388 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
389
201 need (enc, len += 11); 390 need (enc, len += 11);
202 sprintf (enc->cur, "\\u%04x\\u%04x", 391 sprintf (enc->cur, "\\u%04x\\u%04x",
203 (int)((uch - 0x10000) / 0x400 + 0xD800), 392 (int)((uch - 0x10000) / 0x400 + 0xD800),
204 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 393 (int)((uch - 0x10000) % 0x400 + 0xDC00));
205 enc->cur += 12; 394 enc->cur += 12;
206 } 395 }
207 else 396 else
208 { 397 {
209 static char hexdigit [16] = "0123456789abcdef";
210 need (enc, len += 5); 398 need (enc, len += 5);
211 *enc->cur++ = '\\'; 399 *enc->cur++ = '\\';
212 *enc->cur++ = 'u'; 400 *enc->cur++ = 'u';
213 *enc->cur++ = hexdigit [ uch >> 12 ]; 401 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
214 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 402 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
215 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 403 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
216 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 404 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
217 } 405 }
218 406
219 str += clen; 407 str += clen;
220 } 408 }
221 else if (enc->flags & F_LATIN1) 409 else if (enc->json.flags & F_LATIN1)
222 { 410 {
223 *enc->cur++ = uch; 411 *enc->cur++ = uch;
224 str += clen; 412 str += clen;
225 } 413 }
226 else if (is_utf8) 414 else if (is_utf8)
233 while (--clen); 421 while (--clen);
234 } 422 }
235 else 423 else
236 { 424 {
237 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
238 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 426 enc->cur = encode_utf8 (enc->cur, uch);
239 ++str; 427 ++str;
240 } 428 }
241 } 429 }
242 } 430 }
243 } 431 }
244 432
245 --len; 433 --len;
246 } 434 }
247} 435}
248 436
249inline void 437INLINE void
250encode_indent (enc_t *enc) 438encode_indent (enc_t *enc)
251{ 439{
252 if (enc->flags & F_INDENT) 440 if (enc->json.flags & F_INDENT)
253 { 441 {
254 int spaces = enc->indent * INDENT_STEP; 442 int spaces = enc->indent * INDENT_STEP;
255 443
256 need (enc, spaces); 444 need (enc, spaces);
257 memset (enc->cur, ' ', spaces); 445 memset (enc->cur, ' ', spaces);
258 enc->cur += spaces; 446 enc->cur += spaces;
259 } 447 }
260} 448}
261 449
262inline void 450INLINE void
263encode_space (enc_t *enc) 451encode_space (enc_t *enc)
264{ 452{
265 need (enc, 1); 453 need (enc, 1);
266 encode_ch (enc, ' '); 454 encode_ch (enc, ' ');
267} 455}
268 456
269inline void 457INLINE void
270encode_nl (enc_t *enc) 458encode_nl (enc_t *enc)
271{ 459{
272 if (enc->flags & F_INDENT) 460 if (enc->json.flags & F_INDENT)
273 { 461 {
274 need (enc, 1); 462 need (enc, 1);
275 encode_ch (enc, '\n'); 463 encode_ch (enc, '\n');
276 } 464 }
277} 465}
278 466
279inline void 467INLINE void
280encode_comma (enc_t *enc) 468encode_comma (enc_t *enc)
281{ 469{
282 encode_ch (enc, ','); 470 encode_ch (enc, ',');
283 471
284 if (enc->flags & F_INDENT) 472 if (enc->json.flags & F_INDENT)
285 encode_nl (enc); 473 encode_nl (enc);
286 else if (enc->flags & F_SPACE_AFTER) 474 else if (enc->json.flags & F_SPACE_AFTER)
287 encode_space (enc); 475 encode_space (enc);
288} 476}
289 477
290static void encode_sv (enc_t *enc, SV *sv); 478static void encode_sv (enc_t *enc, SV *sv);
291 479
292static void 480static void
293encode_av (enc_t *enc, AV *av) 481encode_av (enc_t *enc, AV *av)
294{ 482{
295 int i, len = av_len (av); 483 int i, len = av_len (av);
296 484
297 if (enc->indent >= enc->maxdepth) 485 if (enc->indent >= enc->json.max_depth)
298 croak ("data structure too deep (hit recursion limit)"); 486 croak (ERR_NESTING_EXCEEDED);
299 487
300 encode_ch (enc, '['); encode_nl (enc); 488 encode_ch (enc, '[');
301 ++enc->indent;
302 489
490 if (len >= 0)
491 {
492 encode_nl (enc); ++enc->indent;
493
303 for (i = 0; i <= len; ++i) 494 for (i = 0; i <= len; ++i)
304 { 495 {
496 SV **svp = av_fetch (av, i, 0);
497
305 encode_indent (enc); 498 encode_indent (enc);
306 encode_sv (enc, *av_fetch (av, i, 0));
307 499
500 if (svp)
501 encode_sv (enc, *svp);
502 else
503 encode_str (enc, "null", 4, 0);
504
308 if (i < len) 505 if (i < len)
309 encode_comma (enc); 506 encode_comma (enc);
310 } 507 }
311 508
509 encode_nl (enc); --enc->indent; encode_indent (enc);
510 }
511
312 encode_nl (enc); 512 encode_ch (enc, ']');
313
314 --enc->indent;
315 encode_indent (enc); encode_ch (enc, ']');
316} 513}
317 514
318static void 515static void
319encode_he (enc_t *enc, HE *he) 516encode_hk (enc_t *enc, HE *he)
320{ 517{
321 encode_ch (enc, '"'); 518 encode_ch (enc, '"');
322 519
323 if (HeKLEN (he) == HEf_SVKEY) 520 if (HeKLEN (he) == HEf_SVKEY)
324 { 521 {
325 SV *sv = HeSVKEY (he); 522 SV *sv = HeSVKEY (he);
326 STRLEN len; 523 STRLEN len;
327 char *str; 524 char *str;
328 525
329 SvGETMAGIC (sv); 526 SvGETMAGIC (sv);
330 str = SvPV (sv, len); 527 str = SvPV (sv, len);
331 528
332 encode_str (enc, str, len, SvUTF8 (sv)); 529 encode_str (enc, str, len, SvUTF8 (sv));
333 } 530 }
334 else 531 else
335 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 532 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
336 533
337 encode_ch (enc, '"'); 534 encode_ch (enc, '"');
338 535
339 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 536 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
340 encode_ch (enc, ':'); 537 encode_ch (enc, ':');
341 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 538 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
342 encode_sv (enc, HeVAL (he));
343} 539}
344 540
345// compare hash entries, used when all keys are bytestrings 541// compare hash entries, used when all keys are bytestrings
346static int 542static int
347he_cmp_fast (const void *a_, const void *b_) 543he_cmp_fast (const void *a_, const void *b_)
352 HE *b = *(HE **)b_; 548 HE *b = *(HE **)b_;
353 549
354 STRLEN la = HeKLEN (a); 550 STRLEN la = HeKLEN (a);
355 STRLEN lb = HeKLEN (b); 551 STRLEN lb = HeKLEN (b);
356 552
357 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 553 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
358 cmp = la - lb; 554 cmp = lb - la;
359 555
360 return cmp; 556 return cmp;
361} 557}
362 558
363// 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
364static int 560static int
365he_cmp_slow (const void *a, const void *b) 561he_cmp_slow (const void *a, const void *b)
366{ 562{
367 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 563 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
368} 564}
369 565
370static void 566static void
371encode_hv (enc_t *enc, HV *hv) 567encode_hv (enc_t *enc, HV *hv)
372{ 568{
373 int count, i; 569 HE *he;
374 570
375 if (enc->indent >= enc->maxdepth) 571 if (enc->indent >= enc->json.max_depth)
376 croak ("data structure too deep (hit recursion limit)"); 572 croak (ERR_NESTING_EXCEEDED);
377 573
378 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 574 encode_ch (enc, '{');
379 575
380 if ((count = hv_iterinit (hv)))
381 {
382 // for canonical output we have to sort by keys first 576 // for canonical output we have to sort by keys first
383 // actually, this is mostly due to the stupid so-called 577 // actually, this is mostly due to the stupid so-called
384 // security workaround added somewhere in 5.8.x. 578 // security workaround added somewhere in 5.8.x
385 // that randomises hash orderings 579 // that randomises hash orderings
386 if (enc->flags & F_CANONICAL) 580 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
581 {
582 int count = hv_iterinit (hv);
583
584 if (SvMAGICAL (hv))
387 { 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 {
388 int fast = 1; 600 int i, fast = 1;
389 HE *he; 601 HE *hes_stack [STACK_HES];
390#if defined(__BORLANDC__) || defined(_MSC_VER) 602 HE **hes = hes_stack;
391 HE **hes = _alloca (count * sizeof (HE)); 603
392#else 604 // allocate larger arrays on the heap
393 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 605 if (count > STACK_HES)
394#endif 606 {
607 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
608 hes = (HE **)SvPVX (sv);
609 }
395 610
396 i = 0; 611 i = 0;
397 while ((he = hv_iternext (hv))) 612 while ((he = hv_iternext (hv)))
398 { 613 {
399 hes [i++] = he; 614 hes [i++] = he;
421 636
422 FREETMPS; 637 FREETMPS;
423 LEAVE; 638 LEAVE;
424 } 639 }
425 640
426 for (i = 0; i < count; ++i) 641 encode_nl (enc); ++enc->indent;
642
643 while (count--)
427 { 644 {
428 encode_indent (enc); 645 encode_indent (enc);
646 he = hes [count];
429 encode_he (enc, hes [i]); 647 encode_hk (enc, he);
648 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
430 649
431 if (i < count - 1) 650 if (count)
432 encode_comma (enc); 651 encode_comma (enc);
433 } 652 }
434 653
435 encode_nl (enc); 654 encode_nl (enc); --enc->indent; encode_indent (enc);
436 } 655 }
656 }
437 else 657 else
658 {
659 if (hv_iterinit (hv) || SvMAGICAL (hv))
660 if ((he = hv_iternext (hv)))
438 { 661 {
439 HE *he = hv_iternext (hv); 662 encode_nl (enc); ++enc->indent;
440 663
441 for (;;) 664 for (;;)
442 { 665 {
443 encode_indent (enc); 666 encode_indent (enc);
444 encode_he (enc, he); 667 encode_hk (enc, he);
668 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
445 669
446 if (!(he = hv_iternext (hv))) 670 if (!(he = hv_iternext (hv)))
447 break; 671 break;
448 672
449 encode_comma (enc); 673 encode_comma (enc);
450 } 674 }
451 675
452 encode_nl (enc); 676 encode_nl (enc); --enc->indent; encode_indent (enc);
453 } 677 }
454 } 678 }
455 679
456 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 680 encode_ch (enc, '}');
457} 681}
458 682
459// encode objects, arrays and special \0=false and \1=true values. 683// encode objects, arrays and special \0=false and \1=true values.
460static void 684static void
461encode_rv (enc_t *enc, SV *sv) 685encode_rv (enc_t *enc, SV *sv)
462{ 686{
463 svtype svt; 687 svtype svt;
688 GV *method;
464 689
465 SvGETMAGIC (sv); 690 SvGETMAGIC (sv);
466 svt = SvTYPE (sv); 691 svt = SvTYPE (sv);
467 692
693 if (expect_false (SvOBJECT (sv)))
694 {
695 HV *boolean_stash = !JSON_SLOW || types_boolean_stash
696 ? types_boolean_stash
697 : gv_stashpv ("Types::Serialiser::Boolean", 1);
698 HV *stash = SvSTASH (sv);
699
700 if (stash == boolean_stash)
701 {
702 if (SvIV (sv))
703 encode_str (enc, "true", 4, 0);
704 else
705 encode_str (enc, "false", 5, 0);
706 }
707 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
708 {
709 int count;
710 dSP;
711
712 ENTER; SAVETMPS; PUSHMARK (SP);
713 EXTEND (SP, 2);
714 // we re-bless the reference to get overload and other niceties right
715 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
716 PUSHs (sv_json);
717
718 PUTBACK;
719 count = call_sv ((SV *)GvCV (method), G_ARRAY);
720 SPAGAIN;
721
722 // catch this surprisingly common error
723 if (SvROK (TOPs) && SvRV (TOPs) == sv)
724 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
725
726 encode_ch (enc, '(');
727 encode_ch (enc, '"');
728 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
729 encode_ch (enc, '"');
730 encode_ch (enc, ')');
731 encode_ch (enc, '[');
732
733 while (count)
734 {
735 encode_sv (enc, SP[1 - count--]);
736
737 if (count)
738 encode_ch (enc, ',');
739 }
740
741 encode_ch (enc, ']');
742
743 PUTBACK;
744
745 FREETMPS; LEAVE;
746 }
747 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
748 {
749 dSP;
750
751 ENTER; SAVETMPS; PUSHMARK (SP);
752 // we re-bless the reference to get overload and other niceties right
753 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
754
755 // calling with G_SCALAR ensures that we always get a 1 return value
756 PUTBACK;
757 call_sv ((SV *)GvCV (method), G_SCALAR);
758 SPAGAIN;
759
760 // catch this surprisingly common error
761 if (SvROK (TOPs) && SvRV (TOPs) == sv)
762 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
763
764 sv = POPs;
765 PUTBACK;
766
767 encode_sv (enc, sv);
768
769 FREETMPS; LEAVE;
770 }
771 else if (enc->json.flags & F_ALLOW_BLESSED)
772 encode_str (enc, "null", 4, 0);
773 else
774 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
775 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
776 }
468 if (svt == SVt_PVHV) 777 else if (svt == SVt_PVHV)
469 encode_hv (enc, (HV *)sv); 778 encode_hv (enc, (HV *)sv);
470 else if (svt == SVt_PVAV) 779 else if (svt == SVt_PVAV)
471 encode_av (enc, (AV *)sv); 780 encode_av (enc, (AV *)sv);
472 else if (svt < SVt_PVAV) 781 else if (svt < SVt_PVAV)
473 { 782 {
474 if (SvNIOK (sv) && SvIV (sv) == 0) 783 STRLEN len = 0;
784 char *pv = svt ? SvPV (sv, len) : 0;
785
786 if (len == 1 && *pv == '1')
787 encode_str (enc, "true", 4, 0);
788 else if (len == 1 && *pv == '0')
475 encode_str (enc, "false", 5, 0); 789 encode_str (enc, "false", 5, 0);
476 else if (SvNIOK (sv) && SvIV (sv) == 1) 790 else if (enc->json.flags & F_ALLOW_UNKNOWN)
477 encode_str (enc, "true", 4, 0); 791 encode_str (enc, "null", 4, 0);
478 else 792 else
479 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 793 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
480 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 794 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481 } 795 }
796 else if (enc->json.flags & F_ALLOW_UNKNOWN)
797 encode_str (enc, "null", 4, 0);
482 else 798 else
483 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 799 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
484 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 800 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
485} 801}
486 802
497 encode_str (enc, str, len, SvUTF8 (sv)); 813 encode_str (enc, str, len, SvUTF8 (sv));
498 encode_ch (enc, '"'); 814 encode_ch (enc, '"');
499 } 815 }
500 else if (SvNOKp (sv)) 816 else if (SvNOKp (sv))
501 { 817 {
818 // trust that perl will do the right thing w.r.t. JSON syntax.
502 need (enc, NV_DIG + 32); 819 need (enc, NV_DIG + 32);
503 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 820 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
504 enc->cur += strlen (enc->cur); 821 enc->cur += strlen (enc->cur);
505 } 822 }
506 else if (SvIOKp (sv)) 823 else if (SvIOKp (sv))
507 { 824 {
508 // we assume we can always read an IV as a UV 825 // we assume we can always read an IV as a UV and vice versa
509 if (SvUV (sv) & ~(UV)0x7fff) 826 // we assume two's complement
827 // we assume no aliasing issues in the union
828 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
829 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
510 { 830 {
831 // optimise the "small number case"
832 // code will likely be branchless and use only a single multiplication
833 // works for numbers up to 59074
834 I32 i = SvIVX (sv);
835 U32 u;
836 char digit, nz = 0;
837
511 need (enc, 32); 838 need (enc, 6);
512 enc->cur += 839
513 SvIsUV(sv) 840 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
514 ? snprintf (enc->cur, 32, "%"UVuf, (UV)SvUVX (sv)) 841 u = i < 0 ? -i : i;
515 : snprintf (enc->cur, 32, "%"IVdf, (IV)SvIVX (sv)); 842
843 // convert to 4.28 fixed-point representation
844 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
845
846 // now output digit by digit, each time masking out the integer part
847 // and multiplying by 5 while moving the decimal point one to the right,
848 // resulting in a net multiplication by 10.
849 // we always write the digit to memory but conditionally increment
850 // the pointer, to enable the use of conditional move instructions.
851 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
852 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
853 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
854 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
855 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
516 } 856 }
517 else 857 else
518 { 858 {
519 // optimise the "small number case" 859 // large integer, use the (rather slow) snprintf way.
520 // code will likely be branchless and use only a single multiplication
521 I32 i = SvIV (sv);
522 U32 u;
523
524 need (enc, 6); 860 need (enc, IVUV_MAXCHARS);
525 861 enc->cur +=
526 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; 862 SvIsUV(sv)
527 u = i < 0 ? -i : i; 863 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
528 864 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
529 // convert to 4.28 fixed-point representation
530 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
531
532 char digit, nz = 0;
533
534 digit = u >> 28; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0xfffffff) * 5;
535 digit = u >> 27; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0x7ffffff) * 5;
536 digit = u >> 26; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0x3ffffff) * 5;
537 digit = u >> 25; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0x1ffffff) * 5;
538 digit = u >> 24; *enc->cur = digit + '0'; nz |= digit; enc->cur += 1;
539 } 865 }
540 } 866 }
541 else if (SvROK (sv)) 867 else if (SvROK (sv))
542 encode_rv (enc, SvRV (sv)); 868 encode_rv (enc, SvRV (sv));
543 else if (!SvOK (sv)) 869 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
544 encode_str (enc, "null", 4, 0); 870 encode_str (enc, "null", 4, 0);
545 else 871 else
546 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 872 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
547 SvPV_nolen (sv), SvFLAGS (sv)); 873 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
548} 874}
549 875
550static SV * 876static SV *
551encode_json (SV *scalar, U32 flags) 877encode_json (SV *scalar, JSON *json)
552{ 878{
553 enc_t enc; 879 enc_t enc;
554 880
555 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 881 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
556 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 882 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
557 883
558 enc.flags = flags; 884 enc.json = *json;
559 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 885 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
560 enc.cur = SvPVX (enc.sv); 886 enc.cur = SvPVX (enc.sv);
561 enc.end = SvEND (enc.sv); 887 enc.end = SvEND (enc.sv);
562 enc.indent = 0; 888 enc.indent = 0;
563 enc.maxdepth = DEC_DEPTH (flags); 889 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
890 : enc.json.flags & F_LATIN1 ? 0x000100UL
891 : 0x110000UL;
564 892
565 SvPOK_only (enc.sv); 893 SvPOK_only (enc.sv);
566 encode_sv (&enc, scalar); 894 encode_sv (&enc, scalar);
895 encode_nl (&enc);
567 896
568 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 897 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
569 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 898 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
570 899
571 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 900 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
572 SvUTF8_on (enc.sv); 901 SvUTF8_on (enc.sv);
573 902
574 if (enc.flags & F_SHRINK) 903 if (enc.json.flags & F_SHRINK)
575 shrink (enc.sv); 904 shrink (enc.sv);
576 905
577 return enc.sv; 906 return enc.sv;
578} 907}
579 908
584typedef struct 913typedef struct
585{ 914{
586 char *cur; // current parser pointer 915 char *cur; // current parser pointer
587 char *end; // end of input string 916 char *end; // end of input string
588 const char *err; // parse error, if != 0 917 const char *err; // parse error, if != 0
589 U32 flags; // F_* 918 JSON json;
590 U32 depth; // recursion depth 919 U32 depth; // recursion depth
591 U32 maxdepth; // recursion depth limit 920 U32 maxdepth; // recursion depth limit
592} dec_t; 921} dec_t;
593 922
594inline void 923INLINE void
924decode_comment (dec_t *dec)
925{
926 // only '#'-style comments allowed a.t.m.
927
928 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
929 ++dec->cur;
930}
931
932INLINE void
595decode_ws (dec_t *dec) 933decode_ws (dec_t *dec)
596{ 934{
597 for (;;) 935 for (;;)
598 { 936 {
599 char ch = *dec->cur; 937 char ch = *dec->cur;
600 938
601 if (ch > 0x20 939 if (ch > 0x20)
940 {
941 if (expect_false (ch == '#'))
942 {
943 if (dec->json.flags & F_RELAXED)
944 decode_comment (dec);
945 else
946 break;
947 }
948 else
949 break;
950 }
602 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 951 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
603 break; 952 break; // parse error, but let higher level handle it, gives better error messages
604 953
605 ++dec->cur; 954 ++dec->cur;
606 } 955 }
607} 956}
608 957
612 if (*dec->cur != ch) \ 961 if (*dec->cur != ch) \
613 ERR (# ch " expected"); \ 962 ERR (# ch " expected"); \
614 ++dec->cur; \ 963 ++dec->cur; \
615 SE 964 SE
616 965
617#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 966#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
618#define DEC_DEC_DEPTH --dec->depth 967#define DEC_DEC_DEPTH --dec->depth
619 968
620static SV *decode_sv (dec_t *dec); 969static SV *decode_sv (dec_t *dec);
621 970
622static signed char decode_hexdigit[256]; 971static signed char decode_hexdigit[256];
625decode_4hex (dec_t *dec) 974decode_4hex (dec_t *dec)
626{ 975{
627 signed char d1, d2, d3, d4; 976 signed char d1, d2, d3, d4;
628 unsigned char *cur = (unsigned char *)dec->cur; 977 unsigned char *cur = (unsigned char *)dec->cur;
629 978
630 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected"); 979 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
631 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected"); 980 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
632 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected"); 981 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
633 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected"); 982 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
634 983
635 dec->cur += 4; 984 dec->cur += 4;
636 985
637 return ((UV)d1) << 12 986 return ((UV)d1) << 12
638 | ((UV)d2) << 8 987 | ((UV)d2) << 8
646static SV * 995static SV *
647decode_str (dec_t *dec) 996decode_str (dec_t *dec)
648{ 997{
649 SV *sv = 0; 998 SV *sv = 0;
650 int utf8 = 0; 999 int utf8 = 0;
1000 char *dec_cur = dec->cur;
651 1001
652 do 1002 do
653 { 1003 {
654 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; 1004 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
655 char *cur = buf; 1005 char *cur = buf;
656 1006
657 do 1007 do
658 { 1008 {
659 unsigned char ch = *(unsigned char *)dec->cur++; 1009 unsigned char ch = *(unsigned char *)dec_cur++;
660 1010
661 if (expect_false (ch == '"')) 1011 if (expect_false (ch == '"'))
662 { 1012 {
663 --dec->cur; 1013 --dec_cur;
664 break; 1014 break;
665 } 1015 }
666 else if (expect_false (ch == '\\')) 1016 else if (expect_false (ch == '\\'))
667 { 1017 {
668 switch (*dec->cur) 1018 switch (*dec_cur)
669 { 1019 {
670 case '\\': 1020 case '\\':
671 case '/': 1021 case '/':
672 case '"': *cur++ = *dec->cur++; break; 1022 case '"': *cur++ = *dec_cur++; break;
673 1023
674 case 'b': ++dec->cur; *cur++ = '\010'; break; 1024 case 'b': ++dec_cur; *cur++ = '\010'; break;
675 case 't': ++dec->cur; *cur++ = '\011'; break; 1025 case 't': ++dec_cur; *cur++ = '\011'; break;
676 case 'n': ++dec->cur; *cur++ = '\012'; break; 1026 case 'n': ++dec_cur; *cur++ = '\012'; break;
677 case 'f': ++dec->cur; *cur++ = '\014'; break; 1027 case 'f': ++dec_cur; *cur++ = '\014'; break;
678 case 'r': ++dec->cur; *cur++ = '\015'; break; 1028 case 'r': ++dec_cur; *cur++ = '\015'; break;
679 1029
680 case 'u': 1030 case 'u':
681 { 1031 {
682 UV lo, hi; 1032 UV lo, hi;
683 ++dec->cur; 1033 ++dec_cur;
684 1034
1035 dec->cur = dec_cur;
685 hi = decode_4hex (dec); 1036 hi = decode_4hex (dec);
1037 dec_cur = dec->cur;
686 if (hi == (UV)-1) 1038 if (hi == (UV)-1)
687 goto fail; 1039 goto fail;
688 1040
689 // possibly a surrogate pair 1041 // possibly a surrogate pair
690 if (hi >= 0xd800) 1042 if (hi >= 0xd800)
691 if (hi < 0xdc00) 1043 if (hi < 0xdc00)
692 { 1044 {
693 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 1045 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
694 ERR ("missing low surrogate character in surrogate pair"); 1046 ERR ("missing low surrogate character in surrogate pair");
695 1047
696 dec->cur += 2; 1048 dec_cur += 2;
697 1049
1050 dec->cur = dec_cur;
698 lo = decode_4hex (dec); 1051 lo = decode_4hex (dec);
1052 dec_cur = dec->cur;
699 if (lo == (UV)-1) 1053 if (lo == (UV)-1)
700 goto fail; 1054 goto fail;
701 1055
702 if (lo < 0xdc00 || lo >= 0xe000) 1056 if (lo < 0xdc00 || lo >= 0xe000)
703 ERR ("surrogate pair expected"); 1057 ERR ("surrogate pair expected");
709 1063
710 if (hi >= 0x80) 1064 if (hi >= 0x80)
711 { 1065 {
712 utf8 = 1; 1066 utf8 = 1;
713 1067
714 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1068 cur = encode_utf8 (cur, hi);
715 } 1069 }
716 else 1070 else
717 *cur++ = hi; 1071 *cur++ = hi;
718 } 1072 }
719 break; 1073 break;
720 1074
721 default: 1075 default:
722 --dec->cur; 1076 --dec_cur;
723 ERR ("illegal backslash escape sequence in string"); 1077 ERR ("illegal backslash escape sequence in string");
724 } 1078 }
725 } 1079 }
726 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1080 else if (expect_true (ch >= 0x20 && ch < 0x80))
727 *cur++ = ch; 1081 *cur++ = ch;
728 else if (ch >= 0x80) 1082 else if (ch >= 0x80)
729 { 1083 {
730 STRLEN clen; 1084 STRLEN clen;
731 UV uch;
732 1085
733 --dec->cur; 1086 --dec_cur;
734 1087
735 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 1088 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
736 if (clen == (STRLEN)-1) 1089 if (clen == (STRLEN)-1)
737 ERR ("malformed UTF-8 character in JSON string"); 1090 ERR ("malformed UTF-8 character in JSON string");
738 1091
739 do 1092 do
740 *cur++ = *dec->cur++; 1093 *cur++ = *dec_cur++;
741 while (--clen); 1094 while (--clen);
742 1095
743 utf8 = 1; 1096 utf8 = 1;
744 } 1097 }
745 else 1098 else
746 { 1099 {
747 --dec->cur; 1100 --dec_cur;
748 1101
749 if (!ch) 1102 if (!ch)
750 ERR ("unexpected end of string while parsing JSON string"); 1103 ERR ("unexpected end of string while parsing JSON string");
751 else 1104 else
752 ERR ("invalid character encountered while parsing JSON string"); 1105 ERR ("invalid character encountered while parsing JSON string");
757 { 1110 {
758 STRLEN len = cur - buf; 1111 STRLEN len = cur - buf;
759 1112
760 if (sv) 1113 if (sv)
761 { 1114 {
762 SvGROW (sv, SvCUR (sv) + len + 1); 1115 STRLEN cur = SvCUR (sv);
1116
1117 if (SvLEN (sv) <= cur + len)
1118 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1119
763 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1120 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
764 SvCUR_set (sv, SvCUR (sv) + len); 1121 SvCUR_set (sv, SvCUR (sv) + len);
765 } 1122 }
766 else 1123 else
767 sv = newSVpvn (buf, len); 1124 sv = newSVpvn (buf, len);
768 } 1125 }
769 } 1126 }
770 while (*dec->cur != '"'); 1127 while (*dec_cur != '"');
771 1128
772 ++dec->cur; 1129 ++dec_cur;
773 1130
774 if (sv) 1131 if (sv)
775 { 1132 {
776 SvPOK_only (sv); 1133 SvPOK_only (sv);
777 *SvEND (sv) = 0; 1134 *SvEND (sv) = 0;
780 SvUTF8_on (sv); 1137 SvUTF8_on (sv);
781 } 1138 }
782 else 1139 else
783 sv = newSVpvn ("", 0); 1140 sv = newSVpvn ("", 0);
784 1141
1142 dec->cur = dec_cur;
785 return sv; 1143 return sv;
786 1144
787fail: 1145fail:
1146 dec->cur = dec_cur;
788 return 0; 1147 return 0;
789} 1148}
790 1149
791static SV * 1150static SV *
792decode_num (dec_t *dec) 1151decode_num (dec_t *dec)
850 is_nv = 1; 1209 is_nv = 1;
851 } 1210 }
852 1211
853 if (!is_nv) 1212 if (!is_nv)
854 { 1213 {
1214 int len = dec->cur - start;
1215
855 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1216 // special case the rather common 1..5-digit-int case
856 if (*start == '-') 1217 if (*start == '-')
857 switch (dec->cur - start) 1218 switch (len)
858 { 1219 {
859 case 2: return newSViv (-( start [1] - '0' )); 1220 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
860 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1221 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
861 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1222 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
862 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1223 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1224 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
863 } 1225 }
864 else 1226 else
865 switch (dec->cur - start) 1227 switch (len)
866 { 1228 {
867 case 1: return newSViv ( start [0] - '0' ); 1229 case 1: return newSViv ( start [0] - '0' * 1);
868 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1230 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
869 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1231 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
870 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1232 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1233 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
871 } 1234 }
872 1235
873 { 1236 {
874 UV uv; 1237 UV uv;
875 int numtype = grok_number (start, dec->cur - start, &uv); 1238 int numtype = grok_number (start, len, &uv);
876 if (numtype & IS_NUMBER_IN_UV) 1239 if (numtype & IS_NUMBER_IN_UV)
877 if (numtype & IS_NUMBER_NEG) 1240 if (numtype & IS_NUMBER_NEG)
878 { 1241 {
879 if (uv < (UV)IV_MIN) 1242 if (uv < (UV)IV_MIN)
880 return newSViv (-(IV)uv); 1243 return newSViv (-(IV)uv);
881 } 1244 }
882 else 1245 else
883 return newSVuv (uv); 1246 return newSVuv (uv);
884 } 1247 }
885 }
886 1248
1249 len -= *start == '-' ? 1 : 0;
1250
1251 // does not fit into IV or UV, try NV
1252 if (len <= NV_DIG)
1253 // fits into NV without loss of precision
1254 return newSVnv (json_atof (start));
1255
1256 // everything else fails, convert it to a string
1257 return newSVpvn (start, dec->cur - start);
1258 }
1259
1260 // loss of precision here
887 return newSVnv (Atof (start)); 1261 return newSVnv (json_atof (start));
888 1262
889fail: 1263fail:
890 return 0; 1264 return 0;
891} 1265}
892 1266
916 if (*dec->cur == ']') 1290 if (*dec->cur == ']')
917 { 1291 {
918 ++dec->cur; 1292 ++dec->cur;
919 break; 1293 break;
920 } 1294 }
921 1295
922 if (*dec->cur != ',') 1296 if (*dec->cur != ',')
923 ERR (", or ] expected while parsing array"); 1297 ERR (", or ] expected while parsing array");
924 1298
925 ++dec->cur; 1299 ++dec->cur;
1300
1301 decode_ws (dec);
1302
1303 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1304 {
1305 ++dec->cur;
1306 break;
1307 }
926 } 1308 }
927 1309
928 DEC_DEC_DEPTH; 1310 DEC_DEC_DEPTH;
929 return newRV_noinc ((SV *)av); 1311 return newRV_noinc ((SV *)av);
930 1312
935} 1317}
936 1318
937static SV * 1319static SV *
938decode_hv (dec_t *dec) 1320decode_hv (dec_t *dec)
939{ 1321{
1322 SV *sv;
940 HV *hv = newHV (); 1323 HV *hv = newHV ();
941 1324
942 DEC_INC_DEPTH; 1325 DEC_INC_DEPTH;
943 decode_ws (dec); 1326 decode_ws (dec);
944 1327
945 if (*dec->cur == '}') 1328 if (*dec->cur == '}')
946 ++dec->cur; 1329 ++dec->cur;
947 else 1330 else
948 for (;;) 1331 for (;;)
949 { 1332 {
1333 EXPECT_CH ('"');
1334
1335 // heuristic: assume that
1336 // a) decode_str + hv_store_ent are abysmally slow.
1337 // b) most hash keys are short, simple ascii text.
1338 // => try to "fast-match" such strings to avoid
1339 // the overhead of decode_str + hv_store_ent.
1340 {
950 SV *key, *value; 1341 SV *value;
1342 char *p = dec->cur;
1343 char *e = p + 24; // only try up to 24 bytes
951 1344
952 decode_ws (dec); EXPECT_CH ('"'); 1345 for (;;)
953
954 key = decode_str (dec);
955 if (!key)
956 goto fail;
957
958 decode_ws (dec); EXPECT_CH (':');
959
960 value = decode_sv (dec);
961 if (!value)
962 { 1346 {
1347 // the >= 0x80 is false on most architectures
1348 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1349 {
1350 // slow path, back up and use decode_str
1351 SV *key = decode_str (dec);
1352 if (!key)
1353 goto fail;
1354
1355 decode_ws (dec); EXPECT_CH (':');
1356
1357 decode_ws (dec);
1358 value = decode_sv (dec);
1359 if (!value)
1360 {
1361 SvREFCNT_dec (key);
1362 goto fail;
1363 }
1364
1365 hv_store_ent (hv, key, value, 0);
963 SvREFCNT_dec (key); 1366 SvREFCNT_dec (key);
1367
1368 break;
1369 }
1370 else if (*p == '"')
1371 {
1372 // fast path, got a simple key
1373 char *key = dec->cur;
1374 int len = p - key;
1375 dec->cur = p + 1;
1376
1377 decode_ws (dec); EXPECT_CH (':');
1378
1379 decode_ws (dec);
1380 value = decode_sv (dec);
1381 if (!value)
964 goto fail; 1382 goto fail;
1383
1384 hv_store (hv, key, len, value, 0);
1385
1386 break;
1387 }
1388
1389 ++p;
965 } 1390 }
966 1391 }
967 hv_store_ent (hv, key, value, 0);
968 SvREFCNT_dec (key);
969 1392
970 decode_ws (dec); 1393 decode_ws (dec);
971 1394
972 if (*dec->cur == '}') 1395 if (*dec->cur == '}')
973 { 1396 {
977 1400
978 if (*dec->cur != ',') 1401 if (*dec->cur != ',')
979 ERR (", or } expected while parsing object/hash"); 1402 ERR (", or } expected while parsing object/hash");
980 1403
981 ++dec->cur; 1404 ++dec->cur;
1405
1406 decode_ws (dec);
1407
1408 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1409 {
1410 ++dec->cur;
1411 break;
1412 }
982 } 1413 }
983 1414
984 DEC_DEC_DEPTH; 1415 DEC_DEC_DEPTH;
985 return newRV_noinc ((SV *)hv); 1416 sv = newRV_noinc ((SV *)hv);
1417
1418 // check filter callbacks
1419 if (dec->json.flags & F_HOOK)
1420 {
1421 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1422 {
1423 HE *cb, *he;
1424
1425 hv_iterinit (hv);
1426 he = hv_iternext (hv);
1427 hv_iterinit (hv);
1428
1429 // the next line creates a mortal sv each time its called.
1430 // might want to optimise this for common cases.
1431 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1432
1433 if (cb)
1434 {
1435 dSP;
1436 int count;
1437
1438 ENTER; SAVETMPS; PUSHMARK (SP);
1439 XPUSHs (HeVAL (he));
1440 sv_2mortal (sv);
1441
1442 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1443
1444 if (count == 1)
1445 {
1446 sv = newSVsv (POPs);
1447 FREETMPS; LEAVE;
1448 return sv;
1449 }
1450
1451 SvREFCNT_inc (sv);
1452 FREETMPS; LEAVE;
1453 }
1454 }
1455
1456 if (dec->json.cb_object)
1457 {
1458 dSP;
1459 int count;
1460
1461 ENTER; SAVETMPS; PUSHMARK (SP);
1462 XPUSHs (sv_2mortal (sv));
1463
1464 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1465
1466 if (count == 1)
1467 {
1468 sv = newSVsv (POPs);
1469 FREETMPS; LEAVE;
1470 return sv;
1471 }
1472
1473 SvREFCNT_inc (sv);
1474 FREETMPS; LEAVE;
1475 }
1476 }
1477
1478 return sv;
986 1479
987fail: 1480fail:
988 SvREFCNT_dec (hv); 1481 SvREFCNT_dec (hv);
989 DEC_DEC_DEPTH; 1482 DEC_DEC_DEPTH;
990 return 0; 1483 return 0;
991} 1484}
992 1485
993static SV * 1486static SV *
1487decode_tag (dec_t *dec)
1488{
1489 SV *tag = 0;
1490 SV *val = 0;
1491
1492 if (!(dec->json.flags & F_ALLOW_TAGS))
1493 ERR ("malformed JSON string, neither array, object, number, string or atom");
1494
1495 ++dec->cur;
1496
1497 decode_ws (dec);
1498
1499 tag = decode_sv (dec);
1500 if (!tag)
1501 goto fail;
1502
1503 if (!SvPOK (tag))
1504 ERR ("malformed JSON string, (tag) must be a string");
1505
1506 decode_ws (dec);
1507
1508 if (*dec->cur != ')')
1509 ERR (") expected after tag");
1510
1511 ++dec->cur;
1512
1513 decode_ws (dec);
1514
1515 val = decode_sv (dec);
1516 if (!val)
1517 goto fail;
1518
1519 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1520 ERR ("malformed JSON string, tag value must be an array");
1521
1522 {
1523 AV *av = (AV *)SvRV (val);
1524 int i, len = av_len (av) + 1;
1525 HV *stash = gv_stashsv (tag, 0);
1526 SV *sv;
1527
1528 if (!stash)
1529 ERR ("cannot decode perl-object (package does not exist)");
1530
1531 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1532
1533 if (!method)
1534 ERR ("cannot decode perl-object (package does not have a THAW method)");
1535
1536 dSP;
1537
1538 ENTER; SAVETMPS; PUSHMARK (SP);
1539 EXTEND (SP, len + 2);
1540 // we re-bless the reference to get overload and other niceties right
1541 PUSHs (tag);
1542 PUSHs (sv_json);
1543
1544 for (i = 0; i < len; ++i)
1545 PUSHs (*av_fetch (av, i, 1));
1546
1547 PUTBACK;
1548 call_sv ((SV *)GvCV (method), G_SCALAR);
1549 SPAGAIN;
1550
1551 SvREFCNT_dec (tag);
1552 SvREFCNT_dec (val);
1553 sv = SvREFCNT_inc (POPs);
1554
1555 PUTBACK;
1556
1557 FREETMPS; LEAVE;
1558
1559 return sv;
1560 }
1561
1562fail:
1563 SvREFCNT_dec (tag);
1564 SvREFCNT_dec (val);
1565 return 0;
1566}
1567
1568static SV *
994decode_sv (dec_t *dec) 1569decode_sv (dec_t *dec)
995{ 1570{
996 decode_ws (dec); 1571 // the beauty of JSON: you need exactly one character lookahead
1572 // to parse everything.
997 switch (*dec->cur) 1573 switch (*dec->cur)
998 { 1574 {
999 case '"': ++dec->cur; return decode_str (dec); 1575 case '"': ++dec->cur; return decode_str (dec);
1000 case '[': ++dec->cur; return decode_av (dec); 1576 case '[': ++dec->cur; return decode_av (dec);
1001 case '{': ++dec->cur; return decode_hv (dec); 1577 case '{': ++dec->cur; return decode_hv (dec);
1578 case '(': return decode_tag (dec);
1002 1579
1003 case '-': 1580 case '-':
1004 case '0': case '1': case '2': case '3': case '4': 1581 case '0': case '1': case '2': case '3': case '4':
1005 case '5': case '6': case '7': case '8': case '9': 1582 case '5': case '6': case '7': case '8': case '9':
1006 return decode_num (dec); 1583 return decode_num (dec);
1007 1584
1008 case 't': 1585 case 't':
1009 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1586 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1010 { 1587 {
1011 dec->cur += 4; 1588 dec->cur += 4;
1589#if JSON_SLOW
1590 types_true = get_bool ("Types::Serialiser::true");
1591#endif
1012 return newSViv (1); 1592 return newSVsv (types_true);
1013 } 1593 }
1014 else 1594 else
1015 ERR ("'true' expected"); 1595 ERR ("'true' expected");
1016 1596
1017 break; 1597 break;
1018 1598
1019 case 'f': 1599 case 'f':
1020 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1600 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1021 { 1601 {
1022 dec->cur += 5; 1602 dec->cur += 5;
1603#if JSON_SLOW
1604 types_false = get_bool ("Types::Serialiser::false");
1605#endif
1023 return newSViv (0); 1606 return newSVsv (types_false);
1024 } 1607 }
1025 else 1608 else
1026 ERR ("'false' expected"); 1609 ERR ("'false' expected");
1027 1610
1028 break; 1611 break;
1037 ERR ("'null' expected"); 1620 ERR ("'null' expected");
1038 1621
1039 break; 1622 break;
1040 1623
1041 default: 1624 default:
1042 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1625 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1043 break; 1626 break;
1044 } 1627 }
1045 1628
1046fail: 1629fail:
1047 return 0; 1630 return 0;
1048} 1631}
1049 1632
1050static SV * 1633static SV *
1051decode_json (SV *string, U32 flags, UV *offset_return) 1634decode_json (SV *string, JSON *json, char **offset_return)
1052{ 1635{
1053 dec_t dec; 1636 dec_t dec;
1054 UV offset;
1055 SV *sv; 1637 SV *sv;
1056 1638
1639 /* work around bugs in 5.10 where manipulating magic values
1640 * makes perl ignore the magic in subsequent accesses.
1641 * also make a copy of non-PV values, to get them into a clean
1642 * state (SvPV should do that, but it's buggy, see below).
1643 */
1057 SvGETMAGIC (string); 1644 /*SvGETMAGIC (string);*/
1645 if (SvMAGICAL (string) || !SvPOK (string))
1646 string = sv_2mortal (newSVsv (string));
1647
1058 SvUPGRADE (string, SVt_PV); 1648 SvUPGRADE (string, SVt_PV);
1059 1649
1060 if (flags & F_UTF8) 1650 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1651 * assertion with -DDEBUGGING, although SvCUR is documented to
1652 * return the xpv_cur field which certainly exists after upgrading.
1653 * according to nicholas clark, calling SvPOK fixes this.
1654 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1655 * and hope for the best.
1656 * Damnit, SvPV_nolen still trips over yet another assertion. This
1657 * assertion business is seriously broken, try yet another workaround
1658 * for the broken -DDEBUGGING.
1659 */
1660 {
1661#ifdef DEBUGGING
1662 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1663#else
1664 STRLEN offset = SvCUR (string);
1665#endif
1666
1667 if (offset > json->max_size && json->max_size)
1668 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1669 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1670 }
1671
1672 if (DECODE_WANTS_OCTETS (json))
1061 sv_utf8_downgrade (string, 0); 1673 sv_utf8_downgrade (string, 0);
1062 else 1674 else
1063 sv_utf8_upgrade (string); 1675 sv_utf8_upgrade (string);
1064 1676
1065 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1677 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1066 1678
1067 dec.flags = flags; 1679 dec.json = *json;
1068 dec.cur = SvPVX (string); 1680 dec.cur = SvPVX (string);
1069 dec.end = SvEND (string); 1681 dec.end = SvEND (string);
1070 dec.err = 0; 1682 dec.err = 0;
1071 dec.depth = 0; 1683 dec.depth = 0;
1072 dec.maxdepth = DEC_DEPTH (dec.flags); 1684
1685 if (dec.json.cb_object || dec.json.cb_sk_object)
1686 dec.json.flags |= F_HOOK;
1073 1687
1074 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1688 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1689
1690 decode_ws (&dec);
1075 sv = decode_sv (&dec); 1691 sv = decode_sv (&dec);
1692
1693 if (offset_return)
1694 *offset_return = dec.cur;
1076 1695
1077 if (!(offset_return || !sv)) 1696 if (!(offset_return || !sv))
1078 { 1697 {
1079 // check for trailing garbage 1698 // check for trailing garbage
1080 decode_ws (&dec); 1699 decode_ws (&dec);
1083 { 1702 {
1084 dec.err = "garbage after JSON object"; 1703 dec.err = "garbage after JSON object";
1085 SvREFCNT_dec (sv); 1704 SvREFCNT_dec (sv);
1086 sv = 0; 1705 sv = 0;
1087 } 1706 }
1088 }
1089
1090 if (offset_return || !sv)
1091 {
1092 offset = dec.flags & F_UTF8
1093 ? dec.cur - SvPVX (string)
1094 : utf8_distance (dec.cur, SvPVX (string));
1095
1096 if (offset_return)
1097 *offset_return = offset;
1098 } 1707 }
1099 1708
1100 if (!sv) 1709 if (!sv)
1101 { 1710 {
1102 SV *uni = sv_newmortal (); 1711 SV *uni = sv_newmortal ();
1108 SAVEVPTR (PL_curcop); 1717 SAVEVPTR (PL_curcop);
1109 PL_curcop = &cop; 1718 PL_curcop = &cop;
1110 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1719 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1111 LEAVE; 1720 LEAVE;
1112 1721
1113 croak ("%s, at character offset %d [\"%s\"]", 1722 croak ("%s, at character offset %d (before \"%s\")",
1114 dec.err, 1723 dec.err,
1115 (int)offset, 1724 (int)ptr_to_index (string, dec.cur),
1116 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1725 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1117 } 1726 }
1118 1727
1119 sv = sv_2mortal (sv); 1728 sv = sv_2mortal (sv);
1120 1729
1121 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1730 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1122 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1731 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1123 1732
1124 return sv; 1733 return sv;
1734}
1735
1736/////////////////////////////////////////////////////////////////////////////
1737// incremental parser
1738
1739static void
1740incr_parse (JSON *self)
1741{
1742 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1743
1744 // the state machine here is a bit convoluted and could be simplified a lot
1745 // but this would make it slower, so...
1746
1747 for (;;)
1748 {
1749 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D
1750 switch (self->incr_mode)
1751 {
1752 // only used for initial whitespace skipping
1753 case INCR_M_WS:
1754 for (;;)
1755 {
1756 if (*p > 0x20)
1757 {
1758 if (*p == '#')
1759 {
1760 self->incr_mode = INCR_M_C0;
1761 goto incr_m_c;
1762 }
1763 else
1764 {
1765 self->incr_mode = INCR_M_JSON;
1766 goto incr_m_json;
1767 }
1768 }
1769 else if (!*p)
1770 goto interrupt;
1771
1772 ++p;
1773 }
1774
1775 // skip a single char inside a string (for \\-processing)
1776 case INCR_M_BS:
1777 if (!*p)
1778 goto interrupt;
1779
1780 ++p;
1781 self->incr_mode = INCR_M_STR;
1782 goto incr_m_str;
1783
1784 // inside #-style comments
1785 case INCR_M_C0:
1786 case INCR_M_C1:
1787 incr_m_c:
1788 for (;;)
1789 {
1790 if (*p == '\n')
1791 {
1792 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1793 break;
1794 }
1795 else if (!*p)
1796 goto interrupt;
1797
1798 ++p;
1799 }
1800
1801 break;
1802
1803 // inside a string
1804 case INCR_M_STR:
1805 incr_m_str:
1806 for (;;)
1807 {
1808 if (*p == '"')
1809 {
1810 ++p;
1811 self->incr_mode = INCR_M_JSON;
1812
1813 if (!self->incr_nest)
1814 goto interrupt;
1815
1816 goto incr_m_json;
1817 }
1818 else if (*p == '\\')
1819 {
1820 ++p; // "virtually" consumes character after \
1821
1822 if (!*p) // if at end of string we have to switch modes
1823 {
1824 self->incr_mode = INCR_M_BS;
1825 goto interrupt;
1826 }
1827 }
1828 else if (!*p)
1829 goto interrupt;
1830
1831 ++p;
1832 }
1833
1834 // after initial ws, outside string
1835 case INCR_M_JSON:
1836 incr_m_json:
1837 for (;;)
1838 {
1839 switch (*p++)
1840 {
1841 case 0:
1842 --p;
1843 goto interrupt;
1844
1845 case 0x09:
1846 case 0x0a:
1847 case 0x0d:
1848 case 0x20:
1849 if (!self->incr_nest)
1850 {
1851 --p; // do not eat the whitespace, let the next round do it
1852 goto interrupt;
1853 }
1854 break;
1855
1856 case '"':
1857 self->incr_mode = INCR_M_STR;
1858 goto incr_m_str;
1859
1860 case '[':
1861 case '{':
1862 case '(':
1863 if (++self->incr_nest > self->max_depth)
1864 croak (ERR_NESTING_EXCEEDED);
1865 break;
1866
1867 case ']':
1868 case '}':
1869 if (--self->incr_nest <= 0)
1870 goto interrupt;
1871 break;
1872
1873 case ')':
1874 --self->incr_nest;
1875 break;
1876
1877 case '#':
1878 self->incr_mode = INCR_M_C1;
1879 goto incr_m_c;
1880 }
1881 }
1882 }
1883
1884 modechange:
1885 ;
1886 }
1887
1888interrupt:
1889 self->incr_pos = p - SvPVX (self->incr_text);
1890 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1891 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1125} 1892}
1126 1893
1127///////////////////////////////////////////////////////////////////////////// 1894/////////////////////////////////////////////////////////////////////////////
1128// XS interface functions 1895// XS interface functions
1129 1896
1130MODULE = JSON::XS PACKAGE = JSON::XS 1897MODULE = JSON::XS PACKAGE = JSON::XS
1131 1898
1132BOOT: 1899BOOT:
1133{ 1900{
1134 int i; 1901 int i;
1135
1136 memset (decode_hexdigit, 0xff, 256);
1137 1902
1138 for (i = 0; i < 256; ++i) 1903 for (i = 0; i < 256; ++i)
1139 decode_hexdigit [i] = 1904 decode_hexdigit [i] =
1140 i >= '0' && i <= '9' ? i - '0' 1905 i >= '0' && i <= '9' ? i - '0'
1141 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1906 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1142 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1907 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1143 : -1; 1908 : -1;
1144 1909
1145 json_stash = gv_stashpv ("JSON::XS", 1); 1910 json_stash = gv_stashpv ("JSON::XS" , 1);
1911 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1912
1913 types_true = get_bool ("Types::Serialiser::true");
1914 types_false = get_bool ("Types::Serialiser::false");
1915
1916 sv_json = newSVpv ("JSON", 0);
1917 SvREADONLY_on (sv_json);
1918
1919 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1146} 1920}
1147 1921
1148PROTOTYPES: DISABLE 1922PROTOTYPES: DISABLE
1149 1923
1150SV *new (char *dummy) 1924void CLONE (...)
1151 CODE: 1925 CODE:
1152 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1926 json_stash = 0;
1927 types_boolean_stash = 0;
1928
1929void new (char *klass)
1930 PPCODE:
1931{
1932 SV *pv = NEWSV (0, sizeof (JSON));
1933 SvPOK_only (pv);
1934 json_init ((JSON *)SvPVX (pv));
1935 XPUSHs (sv_2mortal (sv_bless (
1936 newRV_noinc (pv),
1937 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1938 )));
1939}
1940
1941void ascii (JSON *self, int enable = 1)
1942 ALIAS:
1943 ascii = F_ASCII
1944 latin1 = F_LATIN1
1945 utf8 = F_UTF8
1946 indent = F_INDENT
1947 canonical = F_CANONICAL
1948 space_before = F_SPACE_BEFORE
1949 space_after = F_SPACE_AFTER
1950 pretty = F_PRETTY
1951 allow_nonref = F_ALLOW_NONREF
1952 shrink = F_SHRINK
1953 allow_blessed = F_ALLOW_BLESSED
1954 convert_blessed = F_CONV_BLESSED
1955 relaxed = F_RELAXED
1956 allow_unknown = F_ALLOW_UNKNOWN
1957 allow_tags = F_ALLOW_TAGS
1958 PPCODE:
1959{
1960 if (enable)
1961 self->flags |= ix;
1962 else
1963 self->flags &= ~ix;
1964
1965 XPUSHs (ST (0));
1966}
1967
1968void get_ascii (JSON *self)
1969 ALIAS:
1970 get_ascii = F_ASCII
1971 get_latin1 = F_LATIN1
1972 get_utf8 = F_UTF8
1973 get_indent = F_INDENT
1974 get_canonical = F_CANONICAL
1975 get_space_before = F_SPACE_BEFORE
1976 get_space_after = F_SPACE_AFTER
1977 get_allow_nonref = F_ALLOW_NONREF
1978 get_shrink = F_SHRINK
1979 get_allow_blessed = F_ALLOW_BLESSED
1980 get_convert_blessed = F_CONV_BLESSED
1981 get_relaxed = F_RELAXED
1982 get_allow_unknown = F_ALLOW_UNKNOWN
1983 get_allow_tags = F_ALLOW_TAGS
1984 PPCODE:
1985 XPUSHs (boolSV (self->flags & ix));
1986
1987void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1988 PPCODE:
1989 self->max_depth = max_depth;
1990 XPUSHs (ST (0));
1991
1992U32 get_max_depth (JSON *self)
1993 CODE:
1994 RETVAL = self->max_depth;
1153 OUTPUT: 1995 OUTPUT:
1154 RETVAL 1996 RETVAL
1155 1997
1156SV *ascii (SV *self, int enable = 1) 1998void max_size (JSON *self, U32 max_size = 0)
1157 ALIAS: 1999 PPCODE:
1158 ascii = F_ASCII 2000 self->max_size = max_size;
1159 latin1 = F_LATIN1 2001 XPUSHs (ST (0));
1160 utf8 = F_UTF8 2002
1161 indent = F_INDENT 2003int get_max_size (JSON *self)
1162 canonical = F_CANONICAL
1163 space_before = F_SPACE_BEFORE
1164 space_after = F_SPACE_AFTER
1165 pretty = F_PRETTY
1166 allow_nonref = F_ALLOW_NONREF
1167 shrink = F_SHRINK
1168 CODE: 2004 CODE:
1169{ 2005 RETVAL = self->max_size;
1170 UV *uv = SvJSON (self);
1171 if (enable)
1172 *uv |= ix;
1173 else
1174 *uv &= ~ix;
1175
1176 RETVAL = newSVsv (self);
1177}
1178 OUTPUT: 2006 OUTPUT:
1179 RETVAL 2007 RETVAL
1180 2008
1181SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 2009void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
2010 PPCODE:
2011{
2012 SvREFCNT_dec (self->cb_object);
2013 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
2014
2015 XPUSHs (ST (0));
2016}
2017
2018void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
2019 PPCODE:
2020{
2021 if (!self->cb_sk_object)
2022 self->cb_sk_object = newHV ();
2023
2024 if (SvOK (cb))
2025 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
2026 else
2027 {
2028 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
2029
2030 if (!HvKEYS (self->cb_sk_object))
2031 {
2032 SvREFCNT_dec (self->cb_sk_object);
2033 self->cb_sk_object = 0;
2034 }
2035 }
2036
2037 XPUSHs (ST (0));
2038}
2039
2040void encode (JSON *self, SV *scalar)
2041 PPCODE:
2042 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2043 XPUSHs (scalar);
2044
2045void decode (JSON *self, SV *jsonstr)
2046 PPCODE:
2047 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2048 XPUSHs (jsonstr);
2049
2050void decode_prefix (JSON *self, SV *jsonstr)
2051 PPCODE:
2052{
2053 SV *sv;
2054 char *offset;
2055 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
2056 EXTEND (SP, 2);
2057 PUSHs (sv);
2058 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
2059}
2060
2061void incr_parse (JSON *self, SV *jsonstr = 0)
2062 PPCODE:
2063{
2064 if (!self->incr_text)
2065 self->incr_text = newSVpvn ("", 0);
2066
2067 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2068 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2069 if (DECODE_WANTS_OCTETS (self))
2070 {
2071 if (self->incr_pos)
2072 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2073 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2074
2075 sv_utf8_downgrade (self->incr_text, 0);
2076 }
2077 else
2078 {
2079 sv_utf8_upgrade (self->incr_text);
2080
2081 if (self->incr_pos)
2082 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2083 - (U8 *)SvPVX (self->incr_text);
2084 }
2085
2086 // append data, if any
2087 if (jsonstr)
2088 {
2089 /* make sure both strings have same encoding */
2090 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2091 if (SvUTF8 (jsonstr))
2092 sv_utf8_downgrade (jsonstr, 0);
2093 else
2094 sv_utf8_upgrade (jsonstr);
2095
2096 /* and then just blindly append */
2097 {
2098 STRLEN len;
2099 const char *str = SvPV (jsonstr, len);
2100 STRLEN cur = SvCUR (self->incr_text);
2101
2102 if (SvLEN (self->incr_text) <= cur + len)
2103 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
2104
2105 Move (str, SvEND (self->incr_text), len, char);
2106 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2107 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2108 }
2109 }
2110
2111 if (GIMME_V != G_VOID)
2112 do
2113 {
2114 SV *sv;
2115 char *offset;
2116
2117 if (!INCR_DONE (self))
2118 {
2119 incr_parse (self);
2120
2121 if (self->incr_pos > self->max_size && self->max_size)
2122 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2123 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2124
2125 if (!INCR_DONE (self))
2126 {
2127 // as an optimisation, do not accumulate white space in the incr buffer
2128 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2129 {
2130 self->incr_pos = 0;
2131 SvCUR_set (self->incr_text, 0);
2132 }
2133
2134 break;
2135 }
2136 }
2137
2138 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2139 XPUSHs (sv);
2140
2141 self->incr_pos -= offset - SvPVX (self->incr_text);
2142 self->incr_nest = 0;
2143 self->incr_mode = 0;
2144
2145 sv_chop (self->incr_text, offset);
2146 }
2147 while (GIMME_V == G_ARRAY);
2148}
2149
2150SV *incr_text (JSON *self)
2151 ATTRS: lvalue
1182 CODE: 2152 CODE:
1183{ 2153{
1184 UV *uv = SvJSON (self); 2154 if (self->incr_pos)
1185 UV log2 = 0; 2155 croak ("incr_text can not be called when the incremental parser already started parsing");
1186 2156
1187 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 2157 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1188
1189 while ((1UL << log2) < max_depth)
1190 ++log2;
1191
1192 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1193
1194 RETVAL = newSVsv (self);
1195} 2158}
1196 OUTPUT: 2159 OUTPUT:
1197 RETVAL 2160 RETVAL
1198 2161
2162void incr_skip (JSON *self)
2163 CODE:
2164{
2165 if (self->incr_pos)
2166 {
2167 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2168 self->incr_pos = 0;
2169 self->incr_nest = 0;
2170 self->incr_mode = 0;
2171 }
2172}
2173
2174void incr_reset (JSON *self)
2175 CODE:
2176{
2177 SvREFCNT_dec (self->incr_text);
2178 self->incr_text = 0;
2179 self->incr_pos = 0;
2180 self->incr_nest = 0;
2181 self->incr_mode = 0;
2182}
2183
2184void DESTROY (JSON *self)
2185 CODE:
2186 SvREFCNT_dec (self->cb_sk_object);
2187 SvREFCNT_dec (self->cb_object);
2188 SvREFCNT_dec (self->incr_text);
2189
2190PROTOTYPES: ENABLE
2191
1199void encode (SV *self, SV *scalar) 2192void encode_json (SV *scalar)
1200 PPCODE: 2193 PPCODE:
1201 XPUSHs (encode_json (scalar, *SvJSON (self))); 2194{
2195 JSON json;
2196 json_init (&json);
2197 json.flags |= F_UTF8;
2198 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2199 XPUSHs (scalar);
2200}
1202 2201
1203void decode (SV *self, SV *jsonstr) 2202void decode_json (SV *jsonstr)
1204 PPCODE: 2203 PPCODE:
1205 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1206
1207void decode_prefix (SV *self, SV *jsonstr)
1208 PPCODE:
1209{ 2204{
1210 UV offset; 2205 JSON json;
1211 EXTEND (SP, 2); 2206 json_init (&json);
1212 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 2207 json.flags |= F_UTF8;
1213 PUSHs (sv_2mortal (newSVuv (offset))); 2208 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2209 XPUSHs (jsonstr);
1214} 2210}
1215 2211
1216PROTOTYPES: ENABLE
1217
1218void to_json (SV *scalar)
1219 ALIAS:
1220 objToJson = 0
1221 PPCODE:
1222 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1223
1224void from_json (SV *jsonstr)
1225 ALIAS:
1226 jsonToObj = 0
1227 PPCODE:
1228 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1229

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines