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 1.46 by root, Mon Jun 25 22:11:39 2007 UTC vs.
Revision 1.120 by root, Mon Oct 28 23:23:10 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)
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}
63 116
64///////////////////////////////////////////////////////////////////////////// 117/////////////////////////////////////////////////////////////////////////////
65// utility functions 118// utility functions
66 119
67static UV * 120INLINE SV *
68SvJSON (SV *sv) 121get_bool (const char *name)
69{ 122{
70 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 123 SV *sv = get_sv (name, 1);
71 croak ("object is not of type JSON::XS");
72 124
73 return &SvUVX (SvRV (sv)); 125 SvREADONLY_on (sv);
74} 126 SvREADONLY_on (SvRV (sv));
75 127
76static void 128 return sv;
129}
130
131INLINE void
77shrink (SV *sv) 132shrink (SV *sv)
78{ 133{
79 sv_utf8_downgrade (sv, 1); 134 sv_utf8_downgrade (sv, 1);
135
80 if (SvLEN (sv) > SvCUR (sv) + 1) 136 if (SvLEN (sv) > SvCUR (sv) + 1)
81 { 137 {
82#ifdef SvPV_shrink_to_cur 138#ifdef SvPV_shrink_to_cur
83 SvPV_shrink_to_cur (sv); 139 SvPV_shrink_to_cur (sv);
84#elif defined (SvPV_renew) 140#elif defined (SvPV_renew)
… …
90// 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
91// case of an error. 147// case of an error.
92// we special-case "safe" characters from U+80 .. U+7FF, 148// we special-case "safe" characters from U+80 .. U+7FF,
93// but use the very good perl function to parse anything else. 149// but use the very good perl function to parse anything else.
94// note that we never call this function for a ascii codepoints 150// note that we never call this function for a ascii codepoints
95inline UV 151INLINE UV
96decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 152decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
97{ 153{
98 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 154 if (expect_true (len >= 2
99 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 155 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
100 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 156 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
101 { 157 {
102 *clen = 2; 158 *clen = 2;
103 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 159 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
104 } 160 }
105 else 161 else
106 { 162 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
107 *clen = (STRLEN)-1; 163}
108 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 (;;)
109 } 215 {
110} 216 U8 dig = (U8)*s - '0';
111 217
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}
112///////////////////////////////////////////////////////////////////////////// 293/////////////////////////////////////////////////////////////////////////////
113// encoder 294// encoder
114 295
115// structure used for encoding JSON 296// structure used for encoding JSON
116typedef struct 297typedef struct
117{ 298{
118 char *cur; // SvPVX (sv) + current output position 299 char *cur; // SvPVX (sv) + current output position
119 char *end; // SvEND (sv) 300 char *end; // SvEND (sv)
120 SV *sv; // result scalar 301 SV *sv; // result scalar
121 U32 flags; // F_* 302 JSON json;
122 U32 indent; // indentation level 303 U32 indent; // indentation level
123 U32 maxdepth; // max. indentation/recursion level 304 UV limit; // escape character values >= this value when encoding
124} enc_t; 305} enc_t;
125 306
126inline void 307INLINE void
127need (enc_t *enc, STRLEN len) 308need (enc_t *enc, STRLEN len)
128{ 309{
129 if (expect_false (enc->cur + len >= enc->end)) 310 if (expect_false (enc->cur + len >= enc->end))
130 { 311 {
131 STRLEN cur = enc->cur - SvPVX (enc->sv); 312 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
132 SvGROW (enc->sv, cur + len + 1); 313 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
133 enc->cur = SvPVX (enc->sv) + cur; 314 enc->cur = SvPVX (enc->sv) + cur;
134 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 315 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
135 } 316 }
136} 317}
137 318
138inline void 319INLINE void
139encode_ch (enc_t *enc, char ch) 320encode_ch (enc_t *enc, char ch)
140{ 321{
141 need (enc, 1); 322 need (enc, 1);
142 *enc->cur++ = ch; 323 *enc->cur++ = ch;
143} 324}
… …
197 { 378 {
198 uch = ch; 379 uch = ch;
199 clen = 1; 380 clen = 1;
200 } 381 }
201 382
202 if (uch > 0x10FFFFUL) 383 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
203 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
204
205 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
206 { 384 {
207 if (uch > 0xFFFFUL) 385 if (uch >= 0x10000UL)
208 { 386 {
387 if (uch >= 0x110000UL)
388 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
389
209 need (enc, len += 11); 390 need (enc, len += 11);
210 sprintf (enc->cur, "\\u%04x\\u%04x", 391 sprintf (enc->cur, "\\u%04x\\u%04x",
211 (int)((uch - 0x10000) / 0x400 + 0xD800), 392 (int)((uch - 0x10000) / 0x400 + 0xD800),
212 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 393 (int)((uch - 0x10000) % 0x400 + 0xDC00));
213 enc->cur += 12; 394 enc->cur += 12;
214 } 395 }
215 else 396 else
216 { 397 {
217 static char hexdigit [16] = "0123456789abcdef";
218 need (enc, len += 5); 398 need (enc, len += 5);
219 *enc->cur++ = '\\'; 399 *enc->cur++ = '\\';
220 *enc->cur++ = 'u'; 400 *enc->cur++ = 'u';
221 *enc->cur++ = hexdigit [ uch >> 12 ]; 401 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
222 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 402 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
223 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 403 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
224 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 404 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
225 } 405 }
226 406
227 str += clen; 407 str += clen;
228 } 408 }
229 else if (enc->flags & F_LATIN1) 409 else if (enc->json.flags & F_LATIN1)
230 { 410 {
231 *enc->cur++ = uch; 411 *enc->cur++ = uch;
232 str += clen; 412 str += clen;
233 } 413 }
234 else if (is_utf8) 414 else if (is_utf8)
… …
241 while (--clen); 421 while (--clen);
242 } 422 }
243 else 423 else
244 { 424 {
245 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
246 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 426 enc->cur = encode_utf8 (enc->cur, uch);
247 ++str; 427 ++str;
248 } 428 }
249 } 429 }
250 } 430 }
251 } 431 }
252 432
253 --len; 433 --len;
254 } 434 }
255} 435}
256 436
257inline void 437INLINE void
258encode_indent (enc_t *enc) 438encode_indent (enc_t *enc)
259{ 439{
260 if (enc->flags & F_INDENT) 440 if (enc->json.flags & F_INDENT)
261 { 441 {
262 int spaces = enc->indent * INDENT_STEP; 442 int spaces = enc->indent * INDENT_STEP;
263 443
264 need (enc, spaces); 444 need (enc, spaces);
265 memset (enc->cur, ' ', spaces); 445 memset (enc->cur, ' ', spaces);
266 enc->cur += spaces; 446 enc->cur += spaces;
267 } 447 }
268} 448}
269 449
270inline void 450INLINE void
271encode_space (enc_t *enc) 451encode_space (enc_t *enc)
272{ 452{
273 need (enc, 1); 453 need (enc, 1);
274 encode_ch (enc, ' '); 454 encode_ch (enc, ' ');
275} 455}
276 456
277inline void 457INLINE void
278encode_nl (enc_t *enc) 458encode_nl (enc_t *enc)
279{ 459{
280 if (enc->flags & F_INDENT) 460 if (enc->json.flags & F_INDENT)
281 { 461 {
282 need (enc, 1); 462 need (enc, 1);
283 encode_ch (enc, '\n'); 463 encode_ch (enc, '\n');
284 } 464 }
285} 465}
286 466
287inline void 467INLINE void
288encode_comma (enc_t *enc) 468encode_comma (enc_t *enc)
289{ 469{
290 encode_ch (enc, ','); 470 encode_ch (enc, ',');
291 471
292 if (enc->flags & F_INDENT) 472 if (enc->json.flags & F_INDENT)
293 encode_nl (enc); 473 encode_nl (enc);
294 else if (enc->flags & F_SPACE_AFTER) 474 else if (enc->json.flags & F_SPACE_AFTER)
295 encode_space (enc); 475 encode_space (enc);
296} 476}
297 477
298static void encode_sv (enc_t *enc, SV *sv); 478static void encode_sv (enc_t *enc, SV *sv);
299 479
300static void 480static void
301encode_av (enc_t *enc, AV *av) 481encode_av (enc_t *enc, AV *av)
302{ 482{
303 int i, len = av_len (av); 483 int i, len = av_len (av);
304 484
305 if (enc->indent >= enc->maxdepth) 485 if (enc->indent >= enc->json.max_depth)
306 croak ("data structure too deep (hit recursion limit)"); 486 croak (ERR_NESTING_EXCEEDED);
307 487
308 encode_ch (enc, '['); encode_nl (enc); 488 encode_ch (enc, '[');
309 ++enc->indent;
310 489
490 if (len >= 0)
491 {
492 encode_nl (enc); ++enc->indent;
493
311 for (i = 0; i <= len; ++i) 494 for (i = 0; i <= len; ++i)
312 { 495 {
496 SV **svp = av_fetch (av, i, 0);
497
313 encode_indent (enc); 498 encode_indent (enc);
314 encode_sv (enc, *av_fetch (av, i, 0));
315 499
500 if (svp)
501 encode_sv (enc, *svp);
502 else
503 encode_str (enc, "null", 4, 0);
504
316 if (i < len) 505 if (i < len)
317 encode_comma (enc); 506 encode_comma (enc);
318 } 507 }
319 508
509 encode_nl (enc); --enc->indent; encode_indent (enc);
510 }
511
320 encode_nl (enc); 512 encode_ch (enc, ']');
321
322 --enc->indent;
323 encode_indent (enc); encode_ch (enc, ']');
324} 513}
325 514
326static void 515static void
327encode_he (enc_t *enc, HE *he) 516encode_hk (enc_t *enc, HE *he)
328{ 517{
329 encode_ch (enc, '"'); 518 encode_ch (enc, '"');
330 519
331 if (HeKLEN (he) == HEf_SVKEY) 520 if (HeKLEN (he) == HEf_SVKEY)
332 { 521 {
333 SV *sv = HeSVKEY (he); 522 SV *sv = HeSVKEY (he);
334 STRLEN len; 523 STRLEN len;
335 char *str; 524 char *str;
336 525
337 SvGETMAGIC (sv); 526 SvGETMAGIC (sv);
338 str = SvPV (sv, len); 527 str = SvPV (sv, len);
339 528
340 encode_str (enc, str, len, SvUTF8 (sv)); 529 encode_str (enc, str, len, SvUTF8 (sv));
341 } 530 }
342 else 531 else
343 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 532 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
344 533
345 encode_ch (enc, '"'); 534 encode_ch (enc, '"');
346 535
347 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 536 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
348 encode_ch (enc, ':'); 537 encode_ch (enc, ':');
349 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 538 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
350 encode_sv (enc, HeVAL (he));
351} 539}
352 540
353// compare hash entries, used when all keys are bytestrings 541// compare hash entries, used when all keys are bytestrings
354static int 542static int
355he_cmp_fast (const void *a_, const void *b_) 543he_cmp_fast (const void *a_, const void *b_)
… …
360 HE *b = *(HE **)b_; 548 HE *b = *(HE **)b_;
361 549
362 STRLEN la = HeKLEN (a); 550 STRLEN la = HeKLEN (a);
363 STRLEN lb = HeKLEN (b); 551 STRLEN lb = HeKLEN (b);
364 552
365 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 553 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
366 cmp = la - lb; 554 cmp = lb - la;
367 555
368 return cmp; 556 return cmp;
369} 557}
370 558
371// 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
372static int 560static int
373he_cmp_slow (const void *a, const void *b) 561he_cmp_slow (const void *a, const void *b)
374{ 562{
375 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 563 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
376} 564}
377 565
378static void 566static void
379encode_hv (enc_t *enc, HV *hv) 567encode_hv (enc_t *enc, HV *hv)
380{ 568{
381 int count, i; 569 HE *he;
382 570
383 if (enc->indent >= enc->maxdepth) 571 if (enc->indent >= enc->json.max_depth)
384 croak ("data structure too deep (hit recursion limit)"); 572 croak (ERR_NESTING_EXCEEDED);
385 573
386 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 574 encode_ch (enc, '{');
387 575
388 if ((count = hv_iterinit (hv)))
389 {
390 // for canonical output we have to sort by keys first 576 // for canonical output we have to sort by keys first
391 // actually, this is mostly due to the stupid so-called 577 // actually, this is mostly due to the stupid so-called
392 // security workaround added somewhere in 5.8.x. 578 // security workaround added somewhere in 5.8.x
393 // that randomises hash orderings 579 // that randomises hash orderings
394 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))
395 { 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 {
396 int fast = 1; 600 int i, fast = 1;
397 HE *he; 601 HE *hes_stack [STACK_HES];
398#if defined(__BORLANDC__) || defined(_MSC_VER) 602 HE **hes = hes_stack;
399 HE **hes = _alloca (count * sizeof (HE)); 603
400#else 604 // allocate larger arrays on the heap
401 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 605 if (count > STACK_HES)
402#endif 606 {
607 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
608 hes = (HE **)SvPVX (sv);
609 }
403 610
404 i = 0; 611 i = 0;
405 while ((he = hv_iternext (hv))) 612 while ((he = hv_iternext (hv)))
406 { 613 {
407 hes [i++] = he; 614 hes [i++] = he;
… …
429 636
430 FREETMPS; 637 FREETMPS;
431 LEAVE; 638 LEAVE;
432 } 639 }
433 640
434 for (i = 0; i < count; ++i) 641 encode_nl (enc); ++enc->indent;
642
643 while (count--)
435 { 644 {
436 encode_indent (enc); 645 encode_indent (enc);
646 he = hes [count];
437 encode_he (enc, hes [i]); 647 encode_hk (enc, he);
648 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
438 649
439 if (i < count - 1) 650 if (count)
440 encode_comma (enc); 651 encode_comma (enc);
441 } 652 }
442 653
443 encode_nl (enc); 654 encode_nl (enc); --enc->indent; encode_indent (enc);
444 } 655 }
656 }
445 else 657 else
658 {
659 if (hv_iterinit (hv) || SvMAGICAL (hv))
660 if ((he = hv_iternext (hv)))
446 { 661 {
447 HE *he = hv_iternext (hv); 662 encode_nl (enc); ++enc->indent;
448 663
449 for (;;) 664 for (;;)
450 { 665 {
451 encode_indent (enc); 666 encode_indent (enc);
452 encode_he (enc, he); 667 encode_hk (enc, he);
668 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
453 669
454 if (!(he = hv_iternext (hv))) 670 if (!(he = hv_iternext (hv)))
455 break; 671 break;
456 672
457 encode_comma (enc); 673 encode_comma (enc);
458 } 674 }
459 675
460 encode_nl (enc); 676 encode_nl (enc); --enc->indent; encode_indent (enc);
461 } 677 }
462 } 678 }
463 679
464 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 680 encode_ch (enc, '}');
465} 681}
466 682
467// encode objects, arrays and special \0=false and \1=true values. 683// encode objects, arrays and special \0=false and \1=true values.
468static void 684static void
469encode_rv (enc_t *enc, SV *sv) 685encode_rv (enc_t *enc, SV *sv)
470{ 686{
471 svtype svt; 687 svtype svt;
688 GV *method;
472 689
473 SvGETMAGIC (sv); 690 SvGETMAGIC (sv);
474 svt = SvTYPE (sv); 691 svt = SvTYPE (sv);
475 692
476 if (expect_false (SvOBJECT (sv))) 693 if (expect_false (SvOBJECT (sv)))
477 { 694 {
478 if (SvSTASH (sv) == json_boolean_stash) 695 HV *boolean_stash = !JSON_SLOW || types_boolean_stash
696 ? types_boolean_stash
697 : gv_stashpv ("Types::Serialiser::Boolean", 1);
698 HV *stash = SvSTASH (sv);
699
700 if (stash == boolean_stash)
479 { 701 {
480 if (SvIV (sv) == 0) 702 if (SvIV (sv))
703 encode_str (enc, "true", 4, 0);
704 else
481 encode_str (enc, "false", 5, 0); 705 encode_str (enc, "false", 5, 0);
482 else
483 encode_str (enc, "true", 4, 0);
484 } 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);
485 else 773 else
486 {
487#if 0
488 if (0 && sv_derived_from (rv, "JSON::Literal"))
489 {
490 // not yet
491 }
492#endif
493 if (enc->flags & F_CONV_BLESSED)
494 {
495 // we re-bless the reference to get overload and other niceties right
496 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
497
498 if (to_json)
499 {
500 dSP;
501 ENTER;
502 SAVETMPS;
503 PUSHMARK (SP);
504 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
505
506 // calling with G_SCALAR ensures that we always get a 1 reutrn value
507 // check anyways.
508 PUTBACK;
509 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
510 SPAGAIN;
511
512 encode_sv (enc, POPs);
513
514 FREETMPS;
515 LEAVE;
516 }
517 else if (enc->flags & F_ALLOW_BLESSED)
518 encode_str (enc, "null", 4, 0);
519 else
520 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
521 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
522 }
523 else if (enc->flags & F_ALLOW_BLESSED)
524 encode_str (enc, "null", 4, 0);
525 else
526 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 774 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
527 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 775 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
528 }
529 } 776 }
530 else if (svt == SVt_PVHV) 777 else if (svt == SVt_PVHV)
531 encode_hv (enc, (HV *)sv); 778 encode_hv (enc, (HV *)sv);
532 else if (svt == SVt_PVAV) 779 else if (svt == SVt_PVAV)
533 encode_av (enc, (AV *)sv); 780 encode_av (enc, (AV *)sv);
534 else if (svt < SVt_PVAV) 781 else if (svt < SVt_PVAV)
535 { 782 {
536 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')
537 encode_str (enc, "false", 5, 0); 789 encode_str (enc, "false", 5, 0);
538 else if (SvNIOK (sv) && SvIV (sv) == 1) 790 else if (enc->json.flags & F_ALLOW_UNKNOWN)
539 encode_str (enc, "true", 4, 0); 791 encode_str (enc, "null", 4, 0);
540 else 792 else
541 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",
542 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 794 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
543 } 795 }
796 else if (enc->json.flags & F_ALLOW_UNKNOWN)
797 encode_str (enc, "null", 4, 0);
544 else 798 else
545 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",
546 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 800 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
547} 801}
548 802
… …
566 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 820 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
567 enc->cur += strlen (enc->cur); 821 enc->cur += strlen (enc->cur);
568 } 822 }
569 else if (SvIOKp (sv)) 823 else if (SvIOKp (sv))
570 { 824 {
571 // 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
572 if (SvUV (sv) & ~(UV)0x7fff) 826 // we assume two's complement
573 { 827 // we assume no aliasing issues in the union
574 // large integer, use the (rather slow) snprintf way. 828 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
575 need (enc, sizeof (UV) * 3); 829 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
576 enc->cur +=
577 SvIsUV(sv)
578 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
579 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
580 }
581 else
582 { 830 {
583 // optimise the "small number case" 831 // optimise the "small number case"
584 // code will likely be branchless and use only a single multiplication 832 // code will likely be branchless and use only a single multiplication
833 // works for numbers up to 59074
585 I32 i = SvIV (sv); 834 I32 i = SvIVX (sv);
586 U32 u; 835 U32 u;
587 char digit, nz = 0; 836 char digit, nz = 0;
588 837
589 need (enc, 6); 838 need (enc, 6);
590 839
… …
596 845
597 // now output digit by digit, each time masking out the integer part 846 // now output digit by digit, each time masking out the integer part
598 // and multiplying by 5 while moving the decimal point one to the right, 847 // and multiplying by 5 while moving the decimal point one to the right,
599 // resulting in a net multiplication by 10. 848 // resulting in a net multiplication by 10.
600 // we always write the digit to memory but conditionally increment 849 // we always write the digit to memory but conditionally increment
601 // the pointer, to ease the usage of conditional move instructions. 850 // the pointer, to enable the use of conditional move instructions.
602 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 851 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
603 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 852 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
604 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 853 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
605 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 854 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
606 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 855 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
607 } 856 }
857 else
858 {
859 // large integer, use the (rather slow) snprintf way.
860 need (enc, IVUV_MAXCHARS);
861 enc->cur +=
862 SvIsUV(sv)
863 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
864 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
865 }
608 } 866 }
609 else if (SvROK (sv)) 867 else if (SvROK (sv))
610 encode_rv (enc, SvRV (sv)); 868 encode_rv (enc, SvRV (sv));
611 else if (!SvOK (sv)) 869 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
612 encode_str (enc, "null", 4, 0); 870 encode_str (enc, "null", 4, 0);
613 else 871 else
614 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",
615 SvPV_nolen (sv), SvFLAGS (sv)); 873 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
616} 874}
617 875
618static SV * 876static SV *
619encode_json (SV *scalar, U32 flags) 877encode_json (SV *scalar, JSON *json)
620{ 878{
621 enc_t enc; 879 enc_t enc;
622 880
623 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 881 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
624 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)");
625 883
626 enc.flags = flags; 884 enc.json = *json;
627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 885 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
628 enc.cur = SvPVX (enc.sv); 886 enc.cur = SvPVX (enc.sv);
629 enc.end = SvEND (enc.sv); 887 enc.end = SvEND (enc.sv);
630 enc.indent = 0; 888 enc.indent = 0;
631 enc.maxdepth = DEC_DEPTH (flags); 889 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
890 : enc.json.flags & F_LATIN1 ? 0x000100UL
891 : 0x110000UL;
632 892
633 SvPOK_only (enc.sv); 893 SvPOK_only (enc.sv);
634 encode_sv (&enc, scalar); 894 encode_sv (&enc, scalar);
895 encode_nl (&enc);
635 896
636 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 897 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
637 *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
638 899
639 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 900 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
640 SvUTF8_on (enc.sv); 901 SvUTF8_on (enc.sv);
641 902
642 if (enc.flags & F_SHRINK) 903 if (enc.json.flags & F_SHRINK)
643 shrink (enc.sv); 904 shrink (enc.sv);
644 905
645 return enc.sv; 906 return enc.sv;
646} 907}
647 908
… …
652typedef struct 913typedef struct
653{ 914{
654 char *cur; // current parser pointer 915 char *cur; // current parser pointer
655 char *end; // end of input string 916 char *end; // end of input string
656 const char *err; // parse error, if != 0 917 const char *err; // parse error, if != 0
657 U32 flags; // F_* 918 JSON json;
658 U32 depth; // recursion depth 919 U32 depth; // recursion depth
659 U32 maxdepth; // recursion depth limit 920 U32 maxdepth; // recursion depth limit
660} dec_t; 921} dec_t;
661 922
662inline 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
663decode_ws (dec_t *dec) 933decode_ws (dec_t *dec)
664{ 934{
665 for (;;) 935 for (;;)
666 { 936 {
667 char ch = *dec->cur; 937 char ch = *dec->cur;
668 938
669 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 }
670 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 951 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
671 break; 952 break; // parse error, but let higher level handle it, gives better error messages
672 953
673 ++dec->cur; 954 ++dec->cur;
674 } 955 }
675} 956}
676 957
… …
680 if (*dec->cur != ch) \ 961 if (*dec->cur != ch) \
681 ERR (# ch " expected"); \ 962 ERR (# ch " expected"); \
682 ++dec->cur; \ 963 ++dec->cur; \
683 SE 964 SE
684 965
685#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)
686#define DEC_DEC_DEPTH --dec->depth 967#define DEC_DEC_DEPTH --dec->depth
687 968
688static SV *decode_sv (dec_t *dec); 969static SV *decode_sv (dec_t *dec);
689 970
690static signed char decode_hexdigit[256]; 971static signed char decode_hexdigit[256];
… …
782 1063
783 if (hi >= 0x80) 1064 if (hi >= 0x80)
784 { 1065 {
785 utf8 = 1; 1066 utf8 = 1;
786 1067
787 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1068 cur = encode_utf8 (cur, hi);
788 } 1069 }
789 else 1070 else
790 *cur++ = hi; 1071 *cur++ = hi;
791 } 1072 }
792 break; 1073 break;
… …
794 default: 1075 default:
795 --dec_cur; 1076 --dec_cur;
796 ERR ("illegal backslash escape sequence in string"); 1077 ERR ("illegal backslash escape sequence in string");
797 } 1078 }
798 } 1079 }
799 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1080 else if (expect_true (ch >= 0x20 && ch < 0x80))
800 *cur++ = ch; 1081 *cur++ = ch;
801 else if (ch >= 0x80) 1082 else if (ch >= 0x80)
802 { 1083 {
803 STRLEN clen; 1084 STRLEN clen;
804 UV uch;
805 1085
806 --dec_cur; 1086 --dec_cur;
807 1087
808 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1088 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
809 if (clen == (STRLEN)-1) 1089 if (clen == (STRLEN)-1)
810 ERR ("malformed UTF-8 character in JSON string"); 1090 ERR ("malformed UTF-8 character in JSON string");
811 1091
812 do 1092 do
813 *cur++ = *dec_cur++; 1093 *cur++ = *dec_cur++;
… …
830 { 1110 {
831 STRLEN len = cur - buf; 1111 STRLEN len = cur - buf;
832 1112
833 if (sv) 1113 if (sv)
834 { 1114 {
835 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
836 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1120 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
837 SvCUR_set (sv, SvCUR (sv) + len); 1121 SvCUR_set (sv, SvCUR (sv) + len);
838 } 1122 }
839 else 1123 else
840 sv = newSVpvn (buf, len); 1124 sv = newSVpvn (buf, len);
… …
925 is_nv = 1; 1209 is_nv = 1;
926 } 1210 }
927 1211
928 if (!is_nv) 1212 if (!is_nv)
929 { 1213 {
1214 int len = dec->cur - start;
1215
930 // 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
931 if (*start == '-') 1217 if (*start == '-')
932 switch (dec->cur - start) 1218 switch (len)
933 { 1219 {
934 case 2: return newSViv (-( start [1] - '0' * 1)); 1220 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
935 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1221 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
936 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));
937 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));
938 } 1225 }
939 else 1226 else
940 switch (dec->cur - start) 1227 switch (len)
941 { 1228 {
942 case 1: return newSViv ( start [0] - '0' * 1); 1229 case 1: return newSViv ( start [0] - '0' * 1);
943 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1230 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
944 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);
945 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);
946 } 1234 }
947 1235
948 { 1236 {
949 UV uv; 1237 UV uv;
950 int numtype = grok_number (start, dec->cur - start, &uv); 1238 int numtype = grok_number (start, len, &uv);
951 if (numtype & IS_NUMBER_IN_UV) 1239 if (numtype & IS_NUMBER_IN_UV)
952 if (numtype & IS_NUMBER_NEG) 1240 if (numtype & IS_NUMBER_NEG)
953 { 1241 {
954 if (uv < (UV)IV_MIN) 1242 if (uv < (UV)IV_MIN)
955 return newSViv (-(IV)uv); 1243 return newSViv (-(IV)uv);
956 } 1244 }
957 else 1245 else
958 return newSVuv (uv); 1246 return newSVuv (uv);
959
960 // here would likely be the place for bigint support
961 } 1247 }
962 }
963 1248
964 // if we ever support bigint or bigfloat, this is the place for bigfloat 1249 len -= *start == '-' ? 1 : 0;
1250
1251 // does not fit into IV or UV, try NV
1252 if (len <= NV_DIG)
1253 // fits into NV without loss of precision
1254 return newSVnv (json_atof (start));
1255
1256 // everything else fails, convert it to a string
1257 return newSVpvn (start, dec->cur - start);
1258 }
1259
1260 // loss of precision here
965 return newSVnv (Atof (start)); 1261 return newSVnv (json_atof (start));
966 1262
967fail: 1263fail:
968 return 0; 1264 return 0;
969} 1265}
970 1266
… …
994 if (*dec->cur == ']') 1290 if (*dec->cur == ']')
995 { 1291 {
996 ++dec->cur; 1292 ++dec->cur;
997 break; 1293 break;
998 } 1294 }
999 1295
1000 if (*dec->cur != ',') 1296 if (*dec->cur != ',')
1001 ERR (", or ] expected while parsing array"); 1297 ERR (", or ] expected while parsing array");
1002 1298
1003 ++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 }
1004 } 1308 }
1005 1309
1006 DEC_DEC_DEPTH; 1310 DEC_DEC_DEPTH;
1007 return newRV_noinc ((SV *)av); 1311 return newRV_noinc ((SV *)av);
1008 1312
… …
1013} 1317}
1014 1318
1015static SV * 1319static SV *
1016decode_hv (dec_t *dec) 1320decode_hv (dec_t *dec)
1017{ 1321{
1322 SV *sv;
1018 HV *hv = newHV (); 1323 HV *hv = newHV ();
1019 1324
1020 DEC_INC_DEPTH; 1325 DEC_INC_DEPTH;
1021 decode_ws (dec); 1326 decode_ws (dec);
1022 1327
1023 if (*dec->cur == '}') 1328 if (*dec->cur == '}')
1024 ++dec->cur; 1329 ++dec->cur;
1025 else 1330 else
1026 for (;;) 1331 for (;;)
1027 { 1332 {
1028 decode_ws (dec); EXPECT_CH ('"'); 1333 EXPECT_CH ('"');
1029 1334
1030 // heuristic: assume that 1335 // heuristic: assume that
1031 // a) decode_str + hv_store_ent are abysmally slow 1336 // a) decode_str + hv_store_ent are abysmally slow.
1032 // b) most hash keys are short, simple ascii text 1337 // b) most hash keys are short, simple ascii text.
1033 // so try to "fast-match" such strings to avoid 1338 // => try to "fast-match" such strings to avoid
1034 // the overhead of hv_store_ent. 1339 // the overhead of decode_str + hv_store_ent.
1035 { 1340 {
1036 SV *value; 1341 SV *value;
1037 char *p = dec->cur; 1342 char *p = dec->cur;
1038 char *e = p + 24; // only try up to 24 bytes 1343 char *e = p + 24; // only try up to 24 bytes
1039 1344
1040 for (;;) 1345 for (;;)
1041 { 1346 {
1347 // the >= 0x80 is false on most architectures
1042 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1348 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1043 { 1349 {
1044 // slow path, back up and use decode_str 1350 // slow path, back up and use decode_str
1045 SV *key = decode_str (dec); 1351 SV *key = decode_str (dec);
1046 if (!key) 1352 if (!key)
1047 goto fail; 1353 goto fail;
1048 1354
1049 decode_ws (dec); EXPECT_CH (':'); 1355 decode_ws (dec); EXPECT_CH (':');
1050 1356
1357 decode_ws (dec);
1051 value = decode_sv (dec); 1358 value = decode_sv (dec);
1052 if (!value) 1359 if (!value)
1053 { 1360 {
1054 SvREFCNT_dec (key); 1361 SvREFCNT_dec (key);
1055 goto fail; 1362 goto fail;
… …
1067 int len = p - key; 1374 int len = p - key;
1068 dec->cur = p + 1; 1375 dec->cur = p + 1;
1069 1376
1070 decode_ws (dec); EXPECT_CH (':'); 1377 decode_ws (dec); EXPECT_CH (':');
1071 1378
1379 decode_ws (dec);
1072 value = decode_sv (dec); 1380 value = decode_sv (dec);
1073 if (!value) 1381 if (!value)
1074 goto fail; 1382 goto fail;
1075 1383
1076 hv_store (hv, key, len, value, 0); 1384 hv_store (hv, key, len, value, 0);
… …
1092 1400
1093 if (*dec->cur != ',') 1401 if (*dec->cur != ',')
1094 ERR (", or } expected while parsing object/hash"); 1402 ERR (", or } expected while parsing object/hash");
1095 1403
1096 ++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 }
1097 } 1413 }
1098 1414
1099 DEC_DEC_DEPTH; 1415 DEC_DEC_DEPTH;
1100 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;
1101 1479
1102fail: 1480fail:
1103 SvREFCNT_dec (hv); 1481 SvREFCNT_dec (hv);
1104 DEC_DEC_DEPTH; 1482 DEC_DEC_DEPTH;
1105 return 0; 1483 return 0;
1106} 1484}
1107 1485
1108static 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 tag = decode_sv (dec);
1498 if (!tag)
1499 goto fail;
1500
1501 if (!SvPOK (tag))
1502 ERR ("malformed JSON string, (tag) must be a string");
1503
1504 if (*dec->cur != ')')
1505 ERR (") expected after tag");
1506
1507 ++dec->cur;
1508
1509 val = decode_sv (dec);
1510 if (!val)
1511 goto fail;
1512
1513 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1514 ERR ("malformed JSON string, tag value must be an array");
1515
1516 {
1517 AV *av = (AV *)SvRV (val);
1518 int i, len = av_len (av) + 1;
1519 HV *stash = gv_stashsv (tag, 0);
1520 SV *sv;
1521
1522 if (!stash)
1523 ERR ("cannot decode perl-object (package does not exist)");
1524
1525 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1526
1527 if (!method)
1528 ERR ("cannot decode perl-object (package does not have a THAW method)");
1529
1530 dSP;
1531
1532 ENTER; SAVETMPS; PUSHMARK (SP);
1533 EXTEND (SP, len + 2);
1534 // we re-bless the reference to get overload and other niceties right
1535 PUSHs (tag);
1536 PUSHs (sv_json);
1537
1538 for (i = 0; i < len; ++i)
1539 PUSHs (*av_fetch (av, i, 1));
1540
1541 PUTBACK;
1542 call_sv ((SV *)GvCV (method), G_SCALAR);
1543 SPAGAIN;
1544
1545 SvREFCNT_dec (tag);
1546 SvREFCNT_dec (val);
1547 sv = SvREFCNT_inc (POPs);
1548
1549 PUTBACK;
1550
1551 FREETMPS; LEAVE;
1552
1553 return sv;
1554 }
1555
1556fail:
1557 SvREFCNT_dec (tag);
1558 SvREFCNT_dec (val);
1559 return 0;
1560}
1561
1562static SV *
1109decode_sv (dec_t *dec) 1563decode_sv (dec_t *dec)
1110{ 1564{
1111 decode_ws (dec);
1112
1113 // the beauty of JSON: you need exactly one character lookahead 1565 // the beauty of JSON: you need exactly one character lookahead
1114 // to parse anything. 1566 // to parse everything.
1115 switch (*dec->cur) 1567 switch (*dec->cur)
1116 { 1568 {
1117 case '"': ++dec->cur; return decode_str (dec); 1569 case '"': ++dec->cur; return decode_str (dec);
1118 case '[': ++dec->cur; return decode_av (dec); 1570 case '[': ++dec->cur; return decode_av (dec);
1119 case '{': ++dec->cur; return decode_hv (dec); 1571 case '{': ++dec->cur; return decode_hv (dec);
1572 case '(': return decode_tag (dec);
1120 1573
1121 case '-': 1574 case '-':
1122 case '0': case '1': case '2': case '3': case '4': 1575 case '0': case '1': case '2': case '3': case '4':
1123 case '5': case '6': case '7': case '8': case '9': 1576 case '5': case '6': case '7': case '8': case '9':
1124 return decode_num (dec); 1577 return decode_num (dec);
1125 1578
1126 case 't': 1579 case 't':
1127 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1580 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1128 { 1581 {
1129 dec->cur += 4; 1582 dec->cur += 4;
1130 return SvREFCNT_inc (json_true); 1583#if JSON_SLOW
1584 types_true = get_bool ("Types::Serialiser::true");
1585#endif
1586 return newSVsv (types_true);
1131 } 1587 }
1132 else 1588 else
1133 ERR ("'true' expected"); 1589 ERR ("'true' expected");
1134 1590
1135 break; 1591 break;
1136 1592
1137 case 'f': 1593 case 'f':
1138 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1594 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1139 { 1595 {
1140 dec->cur += 5; 1596 dec->cur += 5;
1141 return SvREFCNT_inc (json_false); 1597#if JSON_SLOW
1598 types_false = get_bool ("Types::Serialiser::false");
1599#endif
1600 return newSVsv (types_false);
1142 } 1601 }
1143 else 1602 else
1144 ERR ("'false' expected"); 1603 ERR ("'false' expected");
1145 1604
1146 break; 1605 break;
… …
1155 ERR ("'null' expected"); 1614 ERR ("'null' expected");
1156 1615
1157 break; 1616 break;
1158 1617
1159 default: 1618 default:
1160 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1619 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1161 break; 1620 break;
1162 } 1621 }
1163 1622
1164fail: 1623fail:
1165 return 0; 1624 return 0;
1166} 1625}
1167 1626
1168static SV * 1627static SV *
1169decode_json (SV *string, U32 flags, UV *offset_return) 1628decode_json (SV *string, JSON *json, char **offset_return)
1170{ 1629{
1171 dec_t dec; 1630 dec_t dec;
1172 UV offset;
1173 SV *sv; 1631 SV *sv;
1174 1632
1633 /* work around bugs in 5.10 where manipulating magic values
1634 * makes perl ignore the magic in subsequent accesses.
1635 * also make a copy of non-PV values, to get them into a clean
1636 * state (SvPV should do that, but it's buggy, see below).
1637 */
1175 SvGETMAGIC (string); 1638 /*SvGETMAGIC (string);*/
1639 if (SvMAGICAL (string) || !SvPOK (string))
1640 string = sv_2mortal (newSVsv (string));
1641
1176 SvUPGRADE (string, SVt_PV); 1642 SvUPGRADE (string, SVt_PV);
1177 1643
1178 if (flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (flags)) 1644 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1645 * assertion with -DDEBUGGING, although SvCUR is documented to
1646 * return the xpv_cur field which certainly exists after upgrading.
1647 * according to nicholas clark, calling SvPOK fixes this.
1648 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1649 * and hope for the best.
1650 * Damnit, SvPV_nolen still trips over yet another assertion. This
1651 * assertion business is seriously broken, try yet another workaround
1652 * for the broken -DDEBUGGING.
1653 */
1654 {
1655#ifdef DEBUGGING
1656 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1657#else
1658 STRLEN offset = SvCUR (string);
1659#endif
1660
1661 if (offset > json->max_size && json->max_size)
1179 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1662 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1180 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (flags)); 1663 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1664 }
1181 1665
1182 if (flags & F_UTF8) 1666 if (DECODE_WANTS_OCTETS (json))
1183 sv_utf8_downgrade (string, 0); 1667 sv_utf8_downgrade (string, 0);
1184 else 1668 else
1185 sv_utf8_upgrade (string); 1669 sv_utf8_upgrade (string);
1186 1670
1187 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1671 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1188 1672
1189 dec.flags = flags; 1673 dec.json = *json;
1190 dec.cur = SvPVX (string); 1674 dec.cur = SvPVX (string);
1191 dec.end = SvEND (string); 1675 dec.end = SvEND (string);
1192 dec.err = 0; 1676 dec.err = 0;
1193 dec.depth = 0; 1677 dec.depth = 0;
1194 dec.maxdepth = DEC_DEPTH (dec.flags); 1678
1679 if (dec.json.cb_object || dec.json.cb_sk_object)
1680 dec.json.flags |= F_HOOK;
1195 1681
1196 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1682 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1683
1684 decode_ws (&dec);
1197 sv = decode_sv (&dec); 1685 sv = decode_sv (&dec);
1686
1687 if (offset_return)
1688 *offset_return = dec.cur;
1198 1689
1199 if (!(offset_return || !sv)) 1690 if (!(offset_return || !sv))
1200 { 1691 {
1201 // check for trailing garbage 1692 // check for trailing garbage
1202 decode_ws (&dec); 1693 decode_ws (&dec);
… …
1205 { 1696 {
1206 dec.err = "garbage after JSON object"; 1697 dec.err = "garbage after JSON object";
1207 SvREFCNT_dec (sv); 1698 SvREFCNT_dec (sv);
1208 sv = 0; 1699 sv = 0;
1209 } 1700 }
1210 }
1211
1212 if (offset_return || !sv)
1213 {
1214 offset = dec.flags & F_UTF8
1215 ? dec.cur - SvPVX (string)
1216 : utf8_distance (dec.cur, SvPVX (string));
1217
1218 if (offset_return)
1219 *offset_return = offset;
1220 } 1701 }
1221 1702
1222 if (!sv) 1703 if (!sv)
1223 { 1704 {
1224 SV *uni = sv_newmortal (); 1705 SV *uni = sv_newmortal ();
… …
1230 SAVEVPTR (PL_curcop); 1711 SAVEVPTR (PL_curcop);
1231 PL_curcop = &cop; 1712 PL_curcop = &cop;
1232 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1713 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1233 LEAVE; 1714 LEAVE;
1234 1715
1235 croak ("%s, at character offset %d [\"%s\"]", 1716 croak ("%s, at character offset %d (before \"%s\")",
1236 dec.err, 1717 dec.err,
1237 (int)offset, 1718 (int)ptr_to_index (string, dec.cur),
1238 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1719 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1239 } 1720 }
1240 1721
1241 sv = sv_2mortal (sv); 1722 sv = sv_2mortal (sv);
1242 1723
1243 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1724 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1244 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1725 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1245 1726
1246 return sv; 1727 return sv;
1728}
1729
1730/////////////////////////////////////////////////////////////////////////////
1731// incremental parser
1732
1733static void
1734incr_parse (JSON *self)
1735{
1736 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1737
1738 // the state machine here is a bit convoluted and could be simplified a lot
1739 // but this would make it slower, so...
1740
1741 for (;;)
1742 {
1743 //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
1744 switch (self->incr_mode)
1745 {
1746 // only used for initial whitespace skipping
1747 case INCR_M_WS:
1748 for (;;)
1749 {
1750 if (*p > 0x20)
1751 {
1752 if (*p == '#')
1753 {
1754 self->incr_mode = INCR_M_C0;
1755 goto incr_m_c;
1756 }
1757 else
1758 {
1759 self->incr_mode = INCR_M_JSON;
1760 goto incr_m_json;
1761 }
1762 }
1763 else if (!*p)
1764 goto interrupt;
1765
1766 ++p;
1767 }
1768
1769 // skip a single char inside a string (for \\-processing)
1770 case INCR_M_BS:
1771 if (!*p)
1772 goto interrupt;
1773
1774 ++p;
1775 self->incr_mode = INCR_M_STR;
1776 goto incr_m_str;
1777
1778 // inside #-style comments
1779 case INCR_M_C0:
1780 case INCR_M_C1:
1781 incr_m_c:
1782 for (;;)
1783 {
1784 if (*p == '\n')
1785 {
1786 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1787 break;
1788 }
1789 else if (!*p)
1790 goto interrupt;
1791
1792 ++p;
1793 }
1794
1795 break;
1796
1797 // inside a string
1798 case INCR_M_STR:
1799 incr_m_str:
1800 for (;;)
1801 {
1802 if (*p == '"')
1803 {
1804 ++p;
1805 self->incr_mode = INCR_M_JSON;
1806
1807 if (!self->incr_nest)
1808 goto interrupt;
1809
1810 goto incr_m_json;
1811 }
1812 else if (*p == '\\')
1813 {
1814 ++p; // "virtually" consumes character after \
1815
1816 if (!*p) // if at end of string we have to switch modes
1817 {
1818 self->incr_mode = INCR_M_BS;
1819 goto interrupt;
1820 }
1821 }
1822 else if (!*p)
1823 goto interrupt;
1824
1825 ++p;
1826 }
1827
1828 // after initial ws, outside string
1829 case INCR_M_JSON:
1830 incr_m_json:
1831 for (;;)
1832 {
1833 switch (*p++)
1834 {
1835 case 0:
1836 --p;
1837 goto interrupt;
1838
1839 case 0x09:
1840 case 0x0a:
1841 case 0x0d:
1842 case 0x20:
1843 if (!self->incr_nest)
1844 {
1845 --p; // do not eat the whitespace, let the next round do it
1846 goto interrupt;
1847 }
1848 break;
1849
1850 case '"':
1851 self->incr_mode = INCR_M_STR;
1852 goto incr_m_str;
1853
1854 case '[':
1855 case '{':
1856 case '(':
1857 if (++self->incr_nest > self->max_depth)
1858 croak (ERR_NESTING_EXCEEDED);
1859 break;
1860
1861 case ']':
1862 case '}':
1863 if (--self->incr_nest <= 0)
1864 goto interrupt;
1865 break;
1866
1867 case ')':
1868 --self->incr_nest;
1869 break;
1870
1871 case '#':
1872 self->incr_mode = INCR_M_C1;
1873 goto incr_m_c;
1874 }
1875 }
1876 }
1877
1878 modechange:
1879 ;
1880 }
1881
1882interrupt:
1883 self->incr_pos = p - SvPVX (self->incr_text);
1884 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1885 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1247} 1886}
1248 1887
1249///////////////////////////////////////////////////////////////////////////// 1888/////////////////////////////////////////////////////////////////////////////
1250// XS interface functions 1889// XS interface functions
1251 1890
… …
1260 i >= '0' && i <= '9' ? i - '0' 1899 i >= '0' && i <= '9' ? i - '0'
1261 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1900 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1262 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1901 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1263 : -1; 1902 : -1;
1264 1903
1265 json_stash = gv_stashpv ("JSON::XS" , 1); 1904 json_stash = gv_stashpv ("JSON::XS" , 1);
1266 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1905 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1267 1906
1268 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1907 types_true = get_bool ("Types::Serialiser::true");
1269 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1908 types_false = get_bool ("Types::Serialiser::false");
1909
1910 sv_json = newSVpv ("JSON", 0);
1911 SvREADONLY_on (sv_json);
1912
1913 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1270} 1914}
1271 1915
1272PROTOTYPES: DISABLE 1916PROTOTYPES: DISABLE
1273 1917
1274SV *new (char *dummy) 1918void CLONE (...)
1275 CODE: 1919 CODE:
1276 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1920 json_stash = 0;
1277 OUTPUT: 1921 types_boolean_stash = 0;
1278 RETVAL
1279 1922
1923void new (char *klass)
1924 PPCODE:
1925{
1926 SV *pv = NEWSV (0, sizeof (JSON));
1927 SvPOK_only (pv);
1928 json_init ((JSON *)SvPVX (pv));
1929 XPUSHs (sv_2mortal (sv_bless (
1930 newRV_noinc (pv),
1931 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1932 )));
1933}
1934
1280SV *ascii (SV *self, int enable = 1) 1935void ascii (JSON *self, int enable = 1)
1281 ALIAS: 1936 ALIAS:
1282 ascii = F_ASCII 1937 ascii = F_ASCII
1283 latin1 = F_LATIN1 1938 latin1 = F_LATIN1
1284 utf8 = F_UTF8 1939 utf8 = F_UTF8
1285 indent = F_INDENT 1940 indent = F_INDENT
… …
1289 pretty = F_PRETTY 1944 pretty = F_PRETTY
1290 allow_nonref = F_ALLOW_NONREF 1945 allow_nonref = F_ALLOW_NONREF
1291 shrink = F_SHRINK 1946 shrink = F_SHRINK
1292 allow_blessed = F_ALLOW_BLESSED 1947 allow_blessed = F_ALLOW_BLESSED
1293 convert_blessed = F_CONV_BLESSED 1948 convert_blessed = F_CONV_BLESSED
1949 relaxed = F_RELAXED
1950 allow_unknown = F_ALLOW_UNKNOWN
1951 allow_tags = F_ALLOW_TAGS
1952 PPCODE:
1953{
1954 if (enable)
1955 self->flags |= ix;
1956 else
1957 self->flags &= ~ix;
1958
1959 XPUSHs (ST (0));
1960}
1961
1962void get_ascii (JSON *self)
1963 ALIAS:
1964 get_ascii = F_ASCII
1965 get_latin1 = F_LATIN1
1966 get_utf8 = F_UTF8
1967 get_indent = F_INDENT
1968 get_canonical = F_CANONICAL
1969 get_space_before = F_SPACE_BEFORE
1970 get_space_after = F_SPACE_AFTER
1971 get_allow_nonref = F_ALLOW_NONREF
1972 get_shrink = F_SHRINK
1973 get_allow_blessed = F_ALLOW_BLESSED
1974 get_convert_blessed = F_CONV_BLESSED
1975 get_relaxed = F_RELAXED
1976 get_allow_unknown = F_ALLOW_UNKNOWN
1977 get_allow_tags = F_ALLOW_TAGS
1978 PPCODE:
1979 XPUSHs (boolSV (self->flags & ix));
1980
1981void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1982 PPCODE:
1983 self->max_depth = max_depth;
1984 XPUSHs (ST (0));
1985
1986U32 get_max_depth (JSON *self)
1294 CODE: 1987 CODE:
1295{ 1988 RETVAL = self->max_depth;
1296 UV *uv = SvJSON (self);
1297 if (enable)
1298 *uv |= ix;
1299 else
1300 *uv &= ~ix;
1301
1302 RETVAL = newSVsv (self);
1303}
1304 OUTPUT: 1989 OUTPUT:
1305 RETVAL 1990 RETVAL
1306 1991
1307SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1992void max_size (JSON *self, U32 max_size = 0)
1993 PPCODE:
1994 self->max_size = max_size;
1995 XPUSHs (ST (0));
1996
1997int get_max_size (JSON *self)
1308 CODE: 1998 CODE:
1309{ 1999 RETVAL = self->max_size;
1310 UV *uv = SvJSON (self);
1311 UV log2 = 0;
1312
1313 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1314
1315 while ((1UL << log2) < max_depth)
1316 ++log2;
1317
1318 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1319
1320 RETVAL = newSVsv (self);
1321}
1322 OUTPUT: 2000 OUTPUT:
1323 RETVAL 2001 RETVAL
1324 2002
1325SV *max_size (SV *self, UV max_size = 0) 2003void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
2004 PPCODE:
2005{
2006 SvREFCNT_dec (self->cb_object);
2007 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
2008
2009 XPUSHs (ST (0));
2010}
2011
2012void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
2013 PPCODE:
2014{
2015 if (!self->cb_sk_object)
2016 self->cb_sk_object = newHV ();
2017
2018 if (SvOK (cb))
2019 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
2020 else
2021 {
2022 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
2023
2024 if (!HvKEYS (self->cb_sk_object))
2025 {
2026 SvREFCNT_dec (self->cb_sk_object);
2027 self->cb_sk_object = 0;
2028 }
2029 }
2030
2031 XPUSHs (ST (0));
2032}
2033
2034void encode (JSON *self, SV *scalar)
2035 PPCODE:
2036 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2037 XPUSHs (scalar);
2038
2039void decode (JSON *self, SV *jsonstr)
2040 PPCODE:
2041 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2042 XPUSHs (jsonstr);
2043
2044void decode_prefix (JSON *self, SV *jsonstr)
2045 PPCODE:
2046{
2047 SV *sv;
2048 char *offset;
2049 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
2050 EXTEND (SP, 2);
2051 PUSHs (sv);
2052 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
2053}
2054
2055void incr_parse (JSON *self, SV *jsonstr = 0)
2056 PPCODE:
2057{
2058 if (!self->incr_text)
2059 self->incr_text = newSVpvn ("", 0);
2060
2061 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2062 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2063 if (DECODE_WANTS_OCTETS (self))
2064 {
2065 if (self->incr_pos)
2066 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2067 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2068
2069 sv_utf8_downgrade (self->incr_text, 0);
2070 }
2071 else
2072 {
2073 sv_utf8_upgrade (self->incr_text);
2074
2075 if (self->incr_pos)
2076 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2077 - (U8 *)SvPVX (self->incr_text);
2078 }
2079
2080 // append data, if any
2081 if (jsonstr)
2082 {
2083 /* make sure both strings have same encoding */
2084 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2085 if (SvUTF8 (jsonstr))
2086 sv_utf8_downgrade (jsonstr, 0);
2087 else
2088 sv_utf8_upgrade (jsonstr);
2089
2090 /* and then just blindly append */
2091 {
2092 STRLEN len;
2093 const char *str = SvPV (jsonstr, len);
2094 STRLEN cur = SvCUR (self->incr_text);
2095
2096 if (SvLEN (self->incr_text) <= cur + len)
2097 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
2098
2099 Move (str, SvEND (self->incr_text), len, char);
2100 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2101 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2102 }
2103 }
2104
2105 if (GIMME_V != G_VOID)
2106 do
2107 {
2108 SV *sv;
2109 char *offset;
2110
2111 if (!INCR_DONE (self))
2112 {
2113 incr_parse (self);
2114
2115 if (self->incr_pos > self->max_size && self->max_size)
2116 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2117 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2118
2119 if (!INCR_DONE (self))
2120 {
2121 // as an optimisation, do not accumulate white space in the incr buffer
2122 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2123 {
2124 self->incr_pos = 0;
2125 SvCUR_set (self->incr_text, 0);
2126 }
2127
2128 break;
2129 }
2130 }
2131
2132 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2133 XPUSHs (sv);
2134
2135 self->incr_pos -= offset - SvPVX (self->incr_text);
2136 self->incr_nest = 0;
2137 self->incr_mode = 0;
2138
2139 sv_chop (self->incr_text, offset);
2140 }
2141 while (GIMME_V == G_ARRAY);
2142}
2143
2144SV *incr_text (JSON *self)
2145 ATTRS: lvalue
1326 CODE: 2146 CODE:
1327{ 2147{
1328 UV *uv = SvJSON (self); 2148 if (self->incr_pos)
1329 UV log2 = 0; 2149 croak ("incr_text can not be called when the incremental parser already started parsing");
1330 2150
1331 if (max_size > 0x80000000UL) max_size = 0x80000000UL; 2151 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1332 if (max_size == 1) max_size = 2;
1333
1334 while ((1UL << log2) < max_size)
1335 ++log2;
1336
1337 *uv = *uv & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1338
1339 RETVAL = newSVsv (self);
1340} 2152}
1341 OUTPUT: 2153 OUTPUT:
1342 RETVAL 2154 RETVAL
1343 2155
2156void incr_skip (JSON *self)
2157 CODE:
2158{
2159 if (self->incr_pos)
2160 {
2161 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2162 self->incr_pos = 0;
2163 self->incr_nest = 0;
2164 self->incr_mode = 0;
2165 }
2166}
2167
2168void incr_reset (JSON *self)
2169 CODE:
2170{
2171 SvREFCNT_dec (self->incr_text);
2172 self->incr_text = 0;
2173 self->incr_pos = 0;
2174 self->incr_nest = 0;
2175 self->incr_mode = 0;
2176}
2177
2178void DESTROY (JSON *self)
2179 CODE:
2180 SvREFCNT_dec (self->cb_sk_object);
2181 SvREFCNT_dec (self->cb_object);
2182 SvREFCNT_dec (self->incr_text);
2183
2184PROTOTYPES: ENABLE
2185
1344void encode (SV *self, SV *scalar) 2186void encode_json (SV *scalar)
1345 PPCODE: 2187 PPCODE:
1346 XPUSHs (encode_json (scalar, *SvJSON (self))); 2188{
2189 JSON json;
2190 json_init (&json);
2191 json.flags |= F_UTF8;
2192 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2193 XPUSHs (scalar);
2194}
1347 2195
1348void decode (SV *self, SV *jsonstr) 2196void decode_json (SV *jsonstr)
1349 PPCODE: 2197 PPCODE:
1350 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1351
1352void decode_prefix (SV *self, SV *jsonstr)
1353 PPCODE:
1354{ 2198{
1355 UV offset; 2199 JSON json;
1356 EXTEND (SP, 2); 2200 json_init (&json);
1357 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 2201 json.flags |= F_UTF8;
1358 PUSHs (sv_2mortal (newSVuv (offset))); 2202 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2203 XPUSHs (jsonstr);
1359} 2204}
1360 2205
1361PROTOTYPES: ENABLE
1362
1363void to_json (SV *scalar)
1364 ALIAS:
1365 objToJson = 0
1366 PPCODE:
1367 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1368
1369void from_json (SV *jsonstr)
1370 ALIAS:
1371 jsonToObj = 0
1372 PPCODE:
1373 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1374

Diff Legend

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